From uli at haase-segel.de Mon Mar 1 01:24:49 1999 From: uli at haase-segel.de (Ulrich =?iso-8859-1?Q?K=F6ster?=) Date: Thu Nov 3 14:41:51 2005 Subject: MacOS X Server v MacOS X References: Message-ID: <36DA5CE0.DA2CEEBC@haase-segel.de> *This message was transferred with a trial version of CommuniGate(tm) Pro* MacOS X will be a "better" (bigger, watered.........) MacOS X but not the end of MacOS X Server. MacOS X includes everything that makes MacOS X Server, in technial terms. In political terms it looks like this: With MacOS X we can expect a "better" MacOS Server. They both decide in one point the market they build for. May be we call this MacOS Server Revision 2. MacOS X Server MacOS X Server Rev.2 MacOS X NetbootServer NetbootServer NetbootClient Server for this Server for this Client for this Server for that Server for that Client for that MaxOS Server is MacOS without the things currently in development. Which are: Carbon for easy porting of traditional MacOS application. Java 2 By the way ... in germany we have the first of march .... it could be very interessting to see which date is currently in Cupertino. Ulrich From Philippe.Robert at uptime.ch Mon Mar 1 01:29:36 1999 From: Philippe.Robert at uptime.ch (Philippe Robert) Date: Thu Nov 3 14:41:51 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <36DA5CE0.DA2CEEBC@haase-segel.de> References: <36DA5CE0.DA2CEEBC@haase-segel.de> Message-ID: <9903010930.AA09477@california.uptime.ch> You wrote: > By the way ... in germany we have the first of march .... it could be very > interessting to see which date is currently in Cupertino. same in switzerland, could there be a bug in NSCalendarDate...*evil grin* sweet dreams, Phil -- Philippe C.D. Robert | Uptime ObjectFactory Inc Unix/OpenStep Software Engineer | http://www.uptime.ch http://www.nice.ch/~phip | info@uptime.ch From Izidor.Jerebic at select-tech.si Mon Mar 1 07:11:16 1999 From: Izidor.Jerebic at select-tech.si (Izidor Jerebic) Date: Thu Nov 3 14:41:51 2005 Subject: [NSFont fon[WithName:matrix:] flips the matrix ??? Message-ID: <199903011510.QAA27461@stsrv.select-tech.si> Hello! When I create fonts with non-standard matrix (not scaled identity matrix), the font's matrix comes out as flipped on PostScript output (screen at least). The element Matrix[3] is negative, but I give strictly positive values. It doesn't happen when using matrix which is identity matrix multiplied by some value. The documentation states that "Fonts created with a matrix other than NSFontIdentityMatrix don't automatically flip themselves in flipped views." Well, I would certainly like to keep the matrix as it was provided to the AppKit, but somehow something there thinks that it must multiply the 3-rd item with -1. Why? I am on OpenStep/Mach 4.2, but am interested in MOXS behaviour and workarounds, too. Currently I just premultiply the 3-rd item with -1 :(. Regards, Izidor Jerebic SELECT Technology From Izidor.Jerebic at select-tech.si Mon Mar 1 07:16:28 1999 From: Izidor.Jerebic at select-tech.si (Izidor Jerebic) Date: Thu Nov 3 14:41:51 2005 Subject: [NSFont fon[WithName:matrix:] flips the matrix ??? In-Reply-To: <199903011510.QAA27461@stsrv.select-tech.si> References: <199903011510.QAA27461@stsrv.select-tech.si> Message-ID: <199903011515.QAA27710@stsrv.select-tech.si> Hello! Sorry to follow up my own post... To clarify: The matrix item M[3] is of course fourth item and not third (indices start with 0)... >When I create fonts with non-standard matrix (not scaled identity >matrix), the font's matrix comes out as flipped on PostScript output >(screen at least). The element Matrix[3] is negative, but I give >strictly positive values. It doesn't happen when using matrix which >is identity matrix multiplied by some value. >The documentation states that "Fonts created with a matrix other >than NSFontIdentityMatrix don't automatically flip themselves in >flipped views." >Well, I would certainly like to keep the matrix as it was provided >to the AppKit, but somehow something there thinks that it must >multiply the 3-rd item with -1. Why? >I am on OpenStep/Mach 4.2, but am interested in MOXS behaviour and >workarounds, too. Currently I just premultiply the 3-rd item with -1 >:(. Regards, Izidor Jerebic SELECT Technology From janos.lobb at yale.edu Mon Mar 1 09:24:41 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:51 2005 Subject: Fwd: Re: MacOS X Server v MacOS X Message-ID: >Date: Mon, 1 Mar 1999 01:30:40 -0800 (PST) >Reply-To: uli@haase-segel.de >Originator: macosx-dev@omnigroup.com >Sender: macosx-dev@omnigroup.com >Precedence: bulk >From: Ulrich K?ster >To: Multiple recipients of list >Subject: Re: MacOS X Server v MacOS X >X-Comment: To unsubscribe, follow directions at >http://www.omnigroup.com/MailArchive/ >MIME-Version: 1.0 >Status: > >*This message was transferred with a trial version of CommuniGate(tm) Pro* >MacOS X will be a "better" (bigger, watered.........) MacOS X but not the >end of MacOS X Server. > >MacOS X includes everything that makes MacOS X Server, in technial terms. > >In political terms it looks like this: With MacOS X we can expect a "better" >MacOS Server. > >They both decide in one point the market they build for. > >May be we call this MacOS Server Revision 2. > >MacOS X Server MacOS X Server Rev.2 MacOS X >NetbootServer NetbootServer NetbootClient >Server for this Server for this Client for this >Server for that Server for that Client for that > >MaxOS Server is MacOS without the things currently in development. Which are: > >Carbon for easy porting of traditional MacOS application. >Java 2 > >By the way ... in germany we have the first of march .... it could be very >interessting to see which date is currently in Cupertino. On the East Coast it is also 3.1.99. But, if I write it in West Europian fashion then it is 1.3.99. I think that can be easily read in the West Coast as 1/3/99, which is equal to January 3rd of 99. That is probably the time in Cupertino.... unless someone took down the boot - I mean MultiBoot - and hitting the table with it with such a passion as Nikita did once - long ago - in the UN. /By the way UN, the clock were stopped there many times just before midnight... / J?nos > >Ulrich > J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From zander at rkinc.com Tue Mar 2 09:22:43 1999 From: zander at rkinc.com (Aleksey Sudakov) Date: Thu Nov 3 14:41:52 2005 Subject: Apache 1.3.4 with DSO on RDR2 Message-ID: <9903021722.AA00532@rkinc.com> Hello, I tried to build apache_1.3.4 with DSO support i.e . with the following configuration options ./configure --enable-module=so --enable-rule=SHARED_CORE and linking did fail. However it worked with ./configure --enable-module=so but linking of any DSO modules failed. I know it's because Apple's shared libraries mechanism is different. I wonder how hard would it be to fix DSO support in Apache. I never figured out how to build dylibs from outside ProjectBuilder. Any ideas, comments? Thanks, Aleksey From pierce at alumni.caltech.edu Tue Mar 2 09:58:04 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:52 2005 Subject: Apache 1.3.4 with DSO on RDR2 In-Reply-To: <9903021722.AA00532@rkinc.com> Message-ID: At 9:34 AM -0800 3/2/99, Aleksey Sudakov wrote: > Hello, > > I tried to build apache_1.3.4 with DSO support i.e . with the following > configuration options > > ./configure --enable-module=so --enable-rule=SHARED_CORE > > and linking did fail. However it worked with > > ./configure --enable-module=so > > but linking of any DSO modules failed. > > I know it's because Apple's shared libraries mechanism is different. I > wonder how hard would it be to fix DSO support in Apache. I never > figured out > how to build dylibs from outside ProjectBuilder. > > Any ideas, comments? > > Thanks, > Aleksey Don't bother, the release has 1.3.4, and uses so. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From zander at rkinc.com Tue Mar 2 14:17:55 1999 From: zander at rkinc.com (Aleksey Sudakov) Date: Thu Nov 3 14:41:52 2005 Subject: EOModeler and CVS Message-ID: <9903022217.AA00762@rkinc.com> Hi, Could somebody confirm or deny that EOModeler on RDR2 remove CVS directories from *.eomodeld the way EOModeler on OpenStep does. If this is the case is there something like OmniEOMoidelerBundle for RDR2? Thanks, Aleksey From wsanchez at apple.com Tue Mar 2 15:02:30 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:52 2005 Subject: Apache 1.3.4 with DSO on RDR2 Message-ID: <199903022302.PAA30894@scv2.apple.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 14851 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990302/628051e7/attachment.bin From daryl at montagetech.com Tue Mar 2 21:28:39 1999 From: daryl at montagetech.com (Daryl Thachuk) Date: Thu Nov 3 14:41:52 2005 Subject: NT Link errors using AIAT Message-ID: <199903030528.WAA03504@cadvision.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 964 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990302/8ff46020/attachment.bin From C.Ridd at isode.com Wed Mar 3 01:02:19 1999 From: C.Ridd at isode.com (Chris Ridd) Date: Thu Nov 3 14:41:52 2005 Subject: Apache 1.3.4 with DSO on RDR2 In-Reply-To: Your message of "Tue, 02 Mar 1999 10:58:04 MST." Message-ID: <15886.920451739@isode.com> On Tue, 02 Mar 1999 10:58:04 MST, "Pierce T. Wetter III" wrote: > At 9:34 AM -0800 3/2/99, Aleksey Sudakov wrote: > > Hello, > > > > I tried to build apache_1.3.4 with DSO support i.e . with the following > > configuration options > Don't bother, the release has 1.3.4, and uses so. > > Pierce You could generalize Alexsey's question a bit by asking how to build an arbitrary Apache configuration from source using additional modules (eg mod_perl) etc. This is important and necessary for many sites. (And of course, how do you get later versions of Apache on your MacOS X box?) In general of course you don't want to rely on a vendor's build of any free software, because it won't be current, and it won't have the latest security fixes applied. Maintaining external packages is extra effort for Apple, which I would prefer in many ways for them to spend on the core OS. The answer is currently to get the modified Apache sources from Apple's ftp site. There's a link to them on the developer web pages. As I understand that Apple's diffs have been applied by the Apache team to the main Apache source tree, hopefully in the future you'll just be able to get the Apache source and (compile and) 'go'. Chris From rcfa at cubiculum.com Tue Mar 2 23:23:46 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903030723.AA07867@kannix.cubiculum.com> > I don't understand the difference between MacOS X Server and MacOS X. > Can someone explain the difference between them and tell why they > should be different products? Why have a separate X Server at all? Among many temporary reasons listed in other replies, because WebObjects will hardly make it into a client version of the OS, and probably neither will the developer tools. So you might say, the Server version is the "professional" version, with developer tools and WebObjects and Unix command line, while the client version will be just that, a client version, possibly w/o any option to access the Unix CLI. Other differences: PPP/POP servers will probably NOT be on the client version, while they will eventually most likely make it into the server edition, etc. The idea of customization for throughput vs. responsiveness makes sense theoretically, but since in fact these are simply a few parameters for the scheduler and disk caching system, they can be set with a NetInfo entry (particularly since a developer system would likely be configured more like a client system with emphasis on responsiveness vs. throughput). Of course, the additional features of the Server version will hike up the price, too. Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From robnewberry at grouplogic.com Wed Mar 3 05:19:22 1999 From: robnewberry at grouplogic.com (Rob Newberry) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903030723.AA07867@kannix.cubiculum.com> Message-ID: >Among many temporary reasons listed in other replies, because >WebObjects will hardly make it into a client version of the OS, Although there have been indications on rumor sites that Apple might include a "Web Objects Lite" for things like remote administration of these client machines via something like personal web sharing. While it is just a rumor, I do think that sounds like a totally cool thing for developers. Just think, if every Mac OS platform had a miniature, limited version of WebObjects, the opportunities for administrating (controlling) the client machine with cool tools by third parties is huge. But Ronald is very, very likely correct that Mac OS X won't have full-blown Web Objects (among other things). Rob --------------------------------------------------------------------- Rob Newberry Director of Fajita Technology Group Logic, Inc. From bbum at codefab.com Wed Mar 3 05:44:37 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: Message-ID: NetInfo provides a total remote administration solution. It scales incredibly well, can be used to administrate just about any aspect of the operating system, and can be easily used to control applications or other user level aspects of the machine. The growing trend throughout the industry of using HTTP to do everything is quite disturbing. HTTP/WWW was designed as a stateless static document publishing system. Anything beyond that is fighting against the feature set of the underlying protocols. It seems that there is a growing population of folks that honestly believe that the Web is the Internet. If WO makes an appearance in Mac OS X Client, it'll likely be the core of a document reference system like it is in the current RDR2 release. b.bum On Wed, 3 Mar 1999, Rob Newberry wrote: > > >Among many temporary reasons listed in other replies, because > >WebObjects will hardly make it into a client version of the OS, > > Although there have been indications on rumor sites that Apple might > include a "Web Objects Lite" for things like remote administration of > these client machines via something like personal web sharing. While it > is just a rumor, I do think that sounds like a totally cool thing for > developers. Just think, if every Mac OS platform had a miniature, limited > version of WebObjects, the opportunities for administrating (controlling) > the client machine with cool tools by third parties is huge. > > But Ronald is very, very likely correct that Mac OS X won't have > full-blown Web Objects (among other things). > > Rob > > --------------------------------------------------------------------- > Rob Newberry > Director of Fajita Technology > Group Logic, Inc. > > > From Malcolm_Crawford at plsys.co.uk Wed Mar 3 05:48:07 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903031348.NAA03627@jasmine.plsys.co.uk> Rob wrote: > Although there have been indications on rumor sites that Apple might > include a "Web Objects Lite" for things like remote administration of > these client machines via something like personal web sharing. While it > is just a rumor, > Umm, not just a rumour, actually; the general point about the potential utility of WebObjects as the basis of a server admin tool was made at the YB BOF at MacWorld by folks from Apple. I wouldn't be surprised to see a 1.0 version in MacOS X -- if Tektronix (amoungst other) laser printers can have their own built-in web sites for configuration, it certainly isn't beyond the wit of the Apple engineers to provide something similar for MacOS X. Indeed, looking at it, NetworkManager.app could quite readily be turned into a web-based application, there's nothing complicated in the interface at all. Best wishes, mmalc. From Malcolm_Crawford at plsys.co.uk Wed Mar 3 05:55:43 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903031355.NAA03634@jasmine.plsys.co.uk> Bill wrote: > NetInfo provides a total remote administration solution. It scales > incredibly well, can be used to administrate just about any aspect of the > operating system, and can be easily used to control applications or other > user level aspects of the machine. > > The growing trend throughout the industry of using HTTP to do everything > is quite disturbing. HTTP/WWW was designed as a stateless static document > publishing system. Anything beyond that is fighting against the feature > set of the underlying protocols. > I thought about this point when I wrote the reply I just sent; I omitted this argument as I thought I'd get the rejoinder that I'm about to send now... ... that I'd anticipate any WO-based tool to meet a perceived need for administration from a non-MacOS-X-based system. If there are tools available to enable sysadmins to do their admining from whatever sys they happen to be sitting at, then this is probably a good thing. *In addition to native tools.* > It seems that there is a growing population of folks that honestly believe > that the Web is the Internet. > This is probably also true. > If WO makes an appearance in Mac OS X Client, it'll likely be the core of > a document reference system like it is in the current RDR2 release. > I'd expect something along the lines of WebObjects 4's WOInfoCenter... Best wishes, mmalc. From david at onestep.co.uk Wed Mar 3 06:08:49 1999 From: david at onestep.co.uk (David Andrew Knight) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X [WO and remote admin] In-Reply-To: References: Message-ID: <9903031408.AA15272@onestep.co.uk> Hi, You wrote: > NetInfo provides a total remote administration solution. It scales > incredibly well, can be used to administrate just about any aspect of the > operating system, and can be easily used to control applications or other > user level aspects of the machine. Whilst I agree that NetInfo is very good this statement goes a little over the top. NetInfo only gives a "total remote administration solution" if you have access to NetInfo enabled system with either the NetInfo GUI client or your are happy to use the CLI tools. If you do not have access to either of these then having a web based front-end to NetInfo is not unreasonable desire, we have Printers with embedded web servers (Xerox and Tek), building control systems, etc. You just need a decent firewall !! --- Regards David Knight OneStep Solutions Plc From rcfa at cubiculum.com Wed Mar 3 06:12:22 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031348.NAA03627@jasmine.plsys.co.uk> References: <199903031348.NAA03627@jasmine.plsys.co.uk> Message-ID: <199903031412.AA09342@kannix.cubiculum.com> you wrote: > Rob wrote: > > Although there have been indications on rumor sites that Apple might > > include a "Web Objects Lite" for things like remote administration of > > these client machines via something like personal web sharing. While it > > is just a rumor, > > > Umm, not just a rumour, actually; the general point about the potential > utility of WebObjects as the basis of a server admin tool was made at the YB > BOF at MacWorld by folks from Apple. I wouldn't be surprised to see a 1.0 > version in MacOS X -- if Tektronix (amoungst other) laser printers can have > their own built-in web sites for configuration, it certainly isn't beyond > the wit of the Apple engineers to provide something similar for MacOS X. > Indeed, looking at it, NetworkManager.app could quite readily be turned into > a web-based application, there's nothing complicated in the interface at > all. And how exactly do you conclude from that that WO would ship with the client? The information, more likely than not, would come from NetInfo and SNMP. Then the network SERVER would provide access to that information to ANY machine by means of a web interface for remote administration. For a standalone machine, the existing NetInfo tools are just fine, and for LAN access, they are better, too, since they don't need to deal with the stateless nature of web transactions. Only for remote monitoring/admin would such a tool be (very) useful, but if you have such a setup, you're more likely than not, running a copy of OSXS on at least one machine on the LAN. I'd be surprised if Apple were to bog down the client with a web server. Accessing information by means of URLs and web apps does not require a web server on the client machine, only a browser, or URL/http-aware Workspace/Finder/OO-API, and these, I bet, we will get. As for the rumored WebObjects-lite: I think that's the 50tpm limit we see in the product that's supposed to be shipping by now but isn't... Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From bbum at codefab.com Wed Mar 3 06:29:52 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X [WO and remote admin] In-Reply-To: <9903031408.AA15272@onestep.co.uk> Message-ID: I should clarify. A web based FRONT END to netinfo would be wonderful [and easy with WO].... however, a web based front end to editing various random system configuration files would be pathetic. As unfortunate of a user interface as the web is, it is convenient. It is too bad that TBL didn't define the original web spec in the context of a user interface protocol-- then again, with DPS and NXHosting or X and XHosting, it was probably not a problem that needed a solution. b.bum From Philippe.Robert at uptime.ch Wed Mar 3 06:39:52 1999 From: Philippe.Robert at uptime.ch (Philippe Robert) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031412.AA09342@kannix.cubiculum.com> References: <199903031412.AA09342@kannix.cubiculum.com> Message-ID: <9903031440.AA19701@california.uptime.ch> You wrote: > As for the rumored WebObjects-lite: I think that's the 50tpm limit > we see in the product that's supposed to be shipping by now but isn't... Why is Apple unable to tell us the reason ... It would not cause any damage, I am sure... sweet dreams, Phil -- Philippe C.D. Robert | Uptime ObjectFactory Inc Unix/OpenStep Software Engineer | http://www.uptime.ch http://www.nice.ch/~phip | info@uptime.ch From lavoie at cst.ca Wed Mar 3 06:57:52 1999 From: lavoie at cst.ca (Martin-Gilles) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X Message-ID: <199903031456.JAA03622@plexus.cst.ca> >You wrote: >> As for the rumored WebObjects-lite: I think that's the 50tpm limit >> we see in the product that's supposed to be shipping by now but isn't... > >Why is Apple unable to tell us the reason ... It would not cause any damage, >I am sure... > I think it's pretty obvious from the demo at MacWorld Tokyo that QuickTime streaming isn't quite...streaming just yet. OSXS ships w/ QT Stream Server (whatever name it'll have). Do the math. Martin-Gilles Lavoie | "No! Try not. Do, or do not. Mac OS programmer, CS&T | There is no try." lavoie@cst.ca | -- Yoda on error handling http://blackhole.cst.ca/ | Y2K statement: hahahaha! From Philippe.Robert at uptime.ch Wed Mar 3 07:08:17 1999 From: Philippe.Robert at uptime.ch (Philippe Robert) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031456.JAA03619@plexus.cst.ca> References: <199903031456.JAA03619@plexus.cst.ca> Message-ID: <9903031509.AA19805@california.uptime.ch> You wrote: > I think it's pretty obvious from the demo at MacWorld Tokyo that QuickTime > streaming isn't quite...streaming just yet. > > OSXS ships w/ QT Stream Server (whatever name it'll have). But why don't they tell us this! So it's just an assumption... sweet dreams, Phil -- Philippe C.D. Robert | Uptime ObjectFactory Inc Unix/OpenStep Software Engineer | http://www.uptime.ch http://www.nice.ch/~phip | info@uptime.ch From Malcolm_Crawford at plsys.co.uk Wed Mar 3 07:15:46 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031412.AA09342@kannix.cubiculum.com> References: <199903031412.AA09342@kannix.cubiculum.com> Message-ID: <199903031515.PAA03667@jasmine.plsys.co.uk> Ronald wrote: > I wrote: > > Umm, not just a rumour, actually; the general point about the potential > > utility of WebObjects as the basis of a server admin tool was made at the > > YB BOF at MacWorld by folks from Apple. I wouldn't be surprised to see a > > 1.0 version in MacOS X -- if Tektronix (amoungst other) laser printers > > can have their own built-in web sites for configuration, it certainly > > isn't beyond the wit of the Apple engineers to provide something similar > > for MacOS X. Indeed, looking at it, NetworkManager.app could quite > > readily be turned into a web-based application, there's nothing > > complicated in the interface at all. > > And how exactly do you conclude from that that WO would ship with > the client? The information, more likely than not, would come > from NetInfo and SNMP. > Umm, why is this so difficult to understand? I'd expect the information to come from NetInfo, just as it does with NetworkManager. It's the *front end* which could readily be translated to the Web using WO -- that being rather the point of it. > Then the network SERVER would provide > access to that information to ANY machine by means of a web interface > for remote administration. > Ah, OK, so it's a reading problem: "the general point about the potential utility of WebObjects as the basis of a *server* admin tool was made at the YB BOF." But, that said... [ > For a standalone machine, the existing NetInfo tools are just fine, > and for LAN access, they are better, too, since they don't need to > deal with the stateless nature of web transactions. > Indeed, that being covered in my followup. ] > Only for remote monitoring/admin would such a tool be (very) useful, > but if you have such a setup, you're more likely than not, running a > copy of OSXS on at least one machine on the LAN. > ... elsewhere (notably re CLI) you have argued for the provision of *choice* and *flexibility* -- why would you now argue against these in this context? As I put in my followup, *some* folks will have a need for remote sysadmin on machines other than ones running MacOS X; if their needs can be catered to easily enough (and I'd guess that converting NetworkManager.app wouldn't be a big job), then why not? > I'd be surprised if Apple were to bog down the client with a web server. > Accessing information by means of URLs and web apps does not require > a web server on the client machine, only a browser, or URL/http-aware > Workspace/Finder/OO-API, and these, I bet, we will get. > Cf above -- I wouldn't expect this to run on every client. This could be an extra on the Server version of MacOS X. Although I'd hardly expect any of these systems to be "bogged down" if they did have a web server running at all times. And it may yet be the case that provision of a service like this would be a useful addition *in some cases*, even for solo 'X machines on a network -- if so, and it's easy to do, why preclude it? Best wishes, mmalc. From lavoie at cst.ca Wed Mar 3 07:22:46 1999 From: lavoie at cst.ca (Martin-Gilles) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X Message-ID: <199903031520.KAA03742@plexus.cst.ca> >You wrote: >> I think it's pretty obvious from the demo at MacWorld Tokyo that QuickTime >> streaming isn't quite...streaming just yet. >> >> OSXS ships w/ QT Stream Server (whatever name it'll have). > >But why don't they tell us this! So it's just an assumption... > >sweet dreams, Phil Politics I guess. If they were to publically admitt that QT bugs is holding off OS X Server, it might make QT 4's entry on the market more difficult. Martin-Gilles Lavoie | "No! Try not. Do, or do not. Mac OS programmer, CS&T | There is no try." lavoie@cst.ca | -- Yoda on error handling http://blackhole.cst.ca/ | Y2K statement: hahahaha! From rcfa at cubiculum.com Wed Mar 3 07:39:21 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031515.PAA03667@jasmine.plsys.co.uk> References: <199903031515.PAA03667@jasmine.plsys.co.uk> Message-ID: <199903031539.AA09802@kannix.cubiculum.com> > ... elsewhere (notably re CLI) you have argued for the provision of > *choice* and *flexibility* -- why would you now argue against these in this > context? As I put in my followup, *some* folks will have a need for remote > sysadmin on machines other than ones running MacOS X; if their needs can be > catered to easily enough (and I'd guess that converting NetworkManager.app > wouldn't be a big job), then why not? I have not argued against such a tool, you misunderstood me here. I only argued against such a tool's server part running on OSX. I hope that it's part of OSXS. The point being, the netinfo server is most likely always going to run OSXS. I could see OSX to only have a client version of NetInfo running (even Xedoc sold such a beast). This would imply, that given that the info needs to come from a NetInfo server, you would run the WO app there as well. The Web client accessing that information can then run on any machine: local or remote, OSX, OSXS, or any other system with a web browser, even a PalmPilot with a web browser connected to the internet via an IrDA link to an Ericsson GSM phone half way on the other side of the planet... Greetings, Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From Malcolm_Crawford at plsys.co.uk Wed Mar 3 07:50:32 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031539.AA09802@kannix.cubiculum.com> References: <199903031515.PAA03667@jasmine.plsys.co.uk> <199903031539.AA09802@kannix.cubiculum.com> Message-ID: <199903031550.PAA03697@jasmine.plsys.co.uk> Ronald wrote: > I have not argued against such a tool, you misunderstood me here. > Touche'. So it looks like we were all actually agreeing with each other all along! Ho de hum. :-) Best wishes, mmalc. From cmh at greendragon.com Wed Mar 3 08:27:54 1999 From: cmh at greendragon.com (Chris Hanson) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031539.AA09802@kannix.cubiculum.com> Message-ID: At 9:51 AM -0600 3/3/99, Ronald C.F. Antony wrote: >I have not argued against such a tool, you misunderstood me here. >I only argued against such a tool's server part running on OSX. Remember that very few people have NetInfo networks right now. A number of people *will* want to set up a network of workstations without a central NetInfo server. (Perhaps they want to put Mac OS X on their engineers' desks, but they don't want to replace the 68040 Performas they're using as file servers with G3s.) What do you do for remote administration then? Do you just say, "Sorry, you're out of luck!" Or do you actually try to meet these peoples' needs? From pierce at alumni.caltech.edu Wed Mar 3 08:42:01 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:52 2005 Subject: Apache 1.3.4 with DSO on RDR2 In-Reply-To: <15886.920451739@isode.com> Message-ID: At 1:09 AM -0800 3/3/99, Chris Ridd wrote: > On Tue, 02 Mar 1999 10:58:04 MST, "Pierce T. Wetter III" wrote: >> At 9:34 AM -0800 3/2/99, Aleksey Sudakov wrote: >> > Hello, >> > >> > I tried to build apache_1.3.4 with DSO support i.e . with the following >> > configuration options >> Don't bother, the release has 1.3.4, and uses so. >> >> Pierce > > You could generalize Alexsey's question a bit by asking how to build an > arbitrary Apache configuration from source using additional modules (eg > mod_perl) etc. This is important and necessary for many sites. (And of > course, how do you get later versions of Apache on your MacOS X box?) > > In general of course you don't want to rely on a vendor's build of any > free software, because it won't be current, and it won't have the > latest security fixes applied. Maintaining external packages is extra > effort for Apple, which I would prefer in many ways for them to spend > on the core OS. Well, Apache isn't a version crazy as most of the other open source sites, so since Mac OS X has 1.3.4, it will be awhile before you have to pull down 1.3.5, and I suspect Apple will provided it anyways, as they're using Apache as their web server. Also, Apache sources are included (if you install them), so you can get at them directly, and I suspect that the diffs have been included back into the Apache distribution, so when 1.3.5 comes out, you can get it from Apache and just build it. As for "using additional modules", since Apache on MacOSX is already using DSO's, you can just include them in the apache.conf file. So I would rather see Aleksey spend his time on some other external package... Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From Malcolm_Crawford at plsys.co.uk Wed Mar 3 08:51:29 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903031651.QAA03745@jasmine.plsys.co.uk> Chris wrote: > At 9:51 AM -0600 3/3/99, Ronald C.F. Antony wrote: > >I have not argued against such a tool, you misunderstood me here. > >I only argued against such a tool's server part running on OSX. > > Remember that very few people have NetInfo networks right now. > > A number of people *will* want to set up a network of workstations without > a central NetInfo server. (Perhaps they want to put Mac OS X on their > engineers' desks, but they don't want to replace the 68040 Performas > they're using as file servers with G3s.) What do you do for remote > administration then? Do you just say, "Sorry, you're out of luck!" Or do > you actually try to meet these peoples' needs? > In what sense "without a central NetInfo server"? You have to have at least one Master NetInfo server. And if the Performas aren't running X then they can't be administered using NetInfo... I started writing a longer answer, but there are too many unknowns in what Apple intends to do to differentiate X from X-Server to address this sensibly. Perhaps this thread should be moved to admin, or talk? Best wishes, mmalc. From pierce at alumni.caltech.edu Wed Mar 3 08:51:45 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: Message-ID: At 5:51 AM -0800 3/3/99, Bill Bumgarner wrote: > NetInfo provides a total remote administration solution. It scales > incredibly well, can be used to administrate just about any aspect of the > operating system, and can be easily used to control applications or other > user level aspects of the machine. > > The growing trend throughout the industry of using HTTP to do everything > is quite disturbing. HTTP/WWW was designed as a stateless static document > publishing system. Anything beyond that is fighting against the feature > set of the underlying protocols. How about a bastard child of NetInfo merged with NIB files? And a client that could deal with that? That would be cool! Actually, I have a way that parts of WebObjects would be cool, for system administration: 99% of the unix services (apache, cron, inittab, daemons, rc, ppp, whatever) are configured by either a text file, or a command line option that could be generated from a text file. Changing those services to use a data file or NetInfo or something editable by a GUI would take forever. This is probably why the "Setup Assistant" in MacOS X really just reads and writes text files for rc to use. So we want to create something where we can have a GUI, but connect it to a text file. Now, what Apple Technology is already based on marking up text files and assigning them to values/objects? WebObjects! WebObjects can take a text file that's been marked up appropriately, along with a description file and turn it into an object graph. Once you have object graphs, you could build a preference panel in IB that lets you wire GUI items up to the objects. Ta da! Instant preference panels! The client program can read a NIB, a text file, a wod, and a woo, and turn it into a text file in the right location. The text file provides a template for the output, the wod describes the components in the text file, and the woo describes the current settings. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From rcfa at cubiculum.com Wed Mar 3 08:52:37 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903031653.AA10162@kannix.cubiculum.com> you wrote: > At 9:51 AM -0600 3/3/99, Ronald C.F. Antony wrote: > >I have not argued against such a tool, you misunderstood me here. > >I only argued against such a tool's server part running on OSX. > > Remember that very few people have NetInfo networks right now. > > A number of people *will* want to set up a network of workstations without > a central NetInfo server. (Perhaps they want to put Mac OS X on their > engineers' desks, but they don't want to replace the 68040 Performas > they're using as file servers with G3s.) What do you do for remote > administration then? Do you just say, "Sorry, you're out of luck!" Or do > you actually try to meet these peoples' needs? Then you pay for your stupidity. If you actually try to administer several machines like that, remotely nonetheless, and you don't tie them together with NetInfo, then you're wasting lots of time and effort. I don't think Apple needs to support moronic system setups. Besides, telnet and CLI tools WILL work, and is good enough for hacks doing a network like this. There are plenty of duties an 040 can take on, but fileserver shouldn't be one of them. Let's be blunt: if Apple doesn't even care about getting OSX(S) to run on all PPC machines, all of which are perfectly adequate to run these OSs, then why would they care about people still using 040s as FILE SERVERS of all things, when OSX-*SERVER* is available and designed to play well with OSX clients? Even more to the point: you administer an entire network, not single machines. How would they administer MacOS machines? The best they currently can do is a sluggish remote display session like with Timbuktu or whatever that stuff is called. People will be able to use such ad-hoc solutions then as now. Apple hasn't supported remote admin in the past, and soon they will, given certain constraints i.e. upgrading software/hardware. Why does that imply they would have to bring comparable solutions to systems they wish they had off their backs rather yesterday than today? I think it's reasonable for Apple to assume that users of OSX clients will use OSXS servers. And it's also reasonable, that users of MOS[78].* might still use MOS[78].* servers. Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From cmh at greendragon.com Wed Mar 3 09:42:21 1999 From: cmh at greendragon.com (Chris Hanson) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031651.QAA03745@jasmine.plsys.co.uk> Message-ID: At 10:59 AM -0600 3/3/99, mmalcolm crawford wrote: >In what sense "without a central NetInfo server"? You have to have at least >one Master NetInfo server. And if the Performas aren't running X then they >can't be administered using NetInfo... I mean NO NETINFO SERVER on the network. Perhaps the shop isn't running their fileservers on Performas... Perhaps they're Linux boxes, or NT machines, or Sun in a NIS+ domain. Perhaps the whole network is just set up so the users can share an Internet connection and a printer or two and there are no servers at all. (You know, like you sometimes see in *school computer labs*?) It's not a "moronic" setup that people should be punished for as Ronald claims; it's the real world, and generally the product of economic decisions. If a school can either buy 30 iMacs, or 26 iMacs and a G3 server, which do you think they'll pick? My point is that there are a lot of situations where users will not be running a NetInfo network, and where you will not be able to convince them to switch to a NetInfo network, and that Mac OS X has to not just *tolerate* such situations, but *work well* with it. Heterogenous networking is the reality, folks, get used to it. >Perhaps this thread should be moved to admin, or talk? Any followups I make after this will be sent to talk. From kc at omnigroup.com Wed Mar 3 09:56:29 1999 From: kc at omnigroup.com (Ken Case) Date: Thu Nov 3 14:41:52 2005 Subject: [Moderator] Re: MacOS X Server v MacOS X Message-ID: <199903031756.JAA27535@ignem.omnigroup.com> As several people have suggested, this discussion thread really belongs on MacOSX-talk, not MacOSX-dev. Thanks, Ken From maury at OAAI.COM Wed Mar 3 10:43:26 1999 From: maury at OAAI.COM (Maury Markowitz) Date: Thu Nov 3 14:41:52 2005 Subject: EPS's vs. TIFF's clipping to userPath's? (2nd try) Message-ID: <199903031834.NAA04490@OAAI.COM> I'm filling shapes with images and having a problem that doesn't seem to make any sense. The shapes have a userPath called path, so I do a... PSgsave(); [path send]; PSclip(); [fill draw]; PSgrestore(); Simple enough. The fills come in a number of varieties, the image fill simply keeps an NSImage and does a... [[image bestRepresentationForDevice:nil] draw]; In it's draw. The problem is that TIFF's work perfectly, but EPS's don't seem to clip and end up drawing outside the path. Is this a known issue? Maury From janos.lobb at yale.edu Wed Mar 3 10:37:07 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903031653.AA10162@kannix.cubiculum.com> Message-ID: At 12:27 PM -0500 3/3/99, Ronald C.F. Antony wrote: >you wrote: >> At 9:51 AM -0600 3/3/99, Ronald C.F. Antony wrote: >> >I have not argued against such a tool, you misunderstood me here. >> >I only argued against such a tool's server part running on OSX. >> >> Remember that very few people have NetInfo networks right now. >> >> A number of people *will* want to set up a network of workstations without >> a central NetInfo server. (Perhaps they want to put Mac OS X on their >> engineers' desks, but they don't want to replace the 68040 Performas >> they're using as file servers with G3s.) What do you do for remote >> administration then? Do you just say, "Sorry, you're out of luck!" Or do >> you actually try to meet these peoples' needs? > >Then you pay for your stupidity. If you actually try to administer several >machines like that, remotely nonetheless, and you don't tie them together >with NetInfo, then you're wasting lots of time and effort. >I don't think Apple needs to support moronic system setups. >Besides, telnet and CLI tools WILL work, and is good enough for hacks >doing a network like this. >There are plenty of duties an 040 can take on, but fileserver shouldn't >be one of them. >Let's be blunt: if Apple doesn't even care about getting OSX(S) to >run on all PPC machines, all of which are perfectly adequate to run >these OSs, then why would they care about people still using 040s as >FILE SERVERS of all things, when OSX-*SERVER* is available and designed >to play well with OSX clients? >Even more to the point: you administer an entire network, not single >machines. How would they administer MacOS machines? The best they >currently can do is a sluggish remote display session like with >Timbuktu or whatever that stuff is called. People will be able to >use such ad-hoc solutions then as now. Apple hasn't supported remote >admin in the past, and soon they will, given certain constraints >i.e. upgrading software/hardware. Why does that imply they would have >to bring comparable solutions to systems they wish they had off their >backs rather yesterday than today? >I think it's reasonable for Apple to assume that users of OSX clients "... Therefore all progress depends on the unreasonable man..." If I let's say look around at Yale where a year ago the Mac as a viable computing platform was almost crashed with a formidable centralized effort, folks who would still love to be on the Mac side of the world can hardly justify and convince central administrators to provide the money for any number of brand new MacOSX Servers even if parts of the software will come deeply discounted. Just politically it will be very difficult and Apple - as of today - does not have matching grants out on the table to counter IBM, HP, Compaq, Dell and ..... Reasonably, no administrator will listen to such unreasonable proposal from any "propeller head". It is a daily fight here to change an LC II for an iMac /which does not run Oracle Application out of box/ to avoid forced conversion to NT. Then I did not got to the 040 line, but just trying to stand on the 60x line for a reasonable hardware for MacOsXS. /"Reasonable" MacOs[78] clients are using Novell and NT servers !! None of them have any reasonable client administration tools worth to mention here/ J?nos P.S. I am running DR2 on a Compaq 486/66 here with 16Mb memory. Convert to a brand new G3 and run MacOsXS "RELEASED", requires WO on PPC. Cannot do that till MacOsXS is not shipped and WO is not on PPC. If I slip my timeframe - next three months - I might not get to MacOsXS at all, because someone might convert me forcefully to an NT machine. Then I cannot cry for even a YB on NT because they will give me Developer 2000 from Oracle or PowerBuilder from Sybase or Delphi from Inprise. >will use OSXS servers. And it's also reasonable, that users of MOS[78].* >might still use MOS[78].* servers. > >Ronald >============================================================================== >"The reasonable man adapts himself to the world; the unreasonable one persists >in trying to adapt the world to himself. Therefore all progress depends on the >unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From zmonster at heliosgroup.com Wed Mar 3 13:17:54 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Compiling C++ and Obj-C Together (in OpenStep NT 4.2)? Message-ID: <5948F7268092D211B2020000D11C0FB8081757@chi-serv2.heliosgroup.com> Hello, I'm having a difficult time trying to compile a framework that has C++ and Obj-C code. There are two types of interaction: - Obj-C objects calling out to C++ objects (i.e. MyClass::DoSomething()) - C++ objects that call out to Obj-C (i.e. [[ObjCClass sharedInstance] doSomething]) (I wrote subclasses of third-party C++ superclasses) I've gotten most of the errors and warnings to go away, but I can't seem to get the last few errors to go away. One of the problems occurs when I import a C++ header shipped with third-party software. I get different errors depending upon the file extension I choose for the Obj-C file. If I choose ".m", I get errors like: "syntax error before "__declspec", where the C++ header looks like this: #ifndef _CONSUMER_HH #define _CONSUMER_HH #ifdef WIN32 #define CLASS_EXPORT __declspec(dllexport) #else #define CLASS_EXPORT #endif class CLASS_EXPORT CTTConsumer { friend class CTTFillFeed; ... If I choose a ".M" extension for my Obj-C class file, I get errors like: "parse error before __stdcall" in the System.framework/Headers/windef.h (and related) files. My questions are: 1. How does one sucessfully mix Obj-C and C++ in a framework? 2. Can you call C++ from an Obj-C class? 3. Can you call Obj-C from a C++ class? 4. What file extension should one choose for the Obj-C class (.m, .M)? 5. What file extension should one choose for the C++ class (.cpp, .C, .cc, etc)? 6. What does the -Obj-C++ compiler C flag do? 7. What does '#pragma cplusplus' do? If necessary, where in source code should this pragma go? 8. Does one have to import special WINNT or C++ headers? Thanks in Advance, Eric From Malcolm_Crawford at plsys.co.uk Wed Mar 3 14:04:35 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: References: Message-ID: <199903032204.WAA03826@jasmine.plsys.co.uk> "dev/talk" bridge -- followups to -talk only please. Chris wrote: > At 10:59 AM -0600 3/3/99, mmalcolm crawford wrote: > >In what sense "without a central NetInfo server"? You have to have at > >least one Master NetInfo server. And if the Performas aren't running X > >then they can't be administered using NetInfo... > > I mean NO NETINFO SERVER on the network. > In which case what's your point?! We're discussing a web UI primarily for *NetInfo* tools... > Perhaps the shop isn't running > their fileservers on Performas... Perhaps they're Linux boxes, or NT > machines, or Sun in a NIS+ domain. Perhaps the whole network is just set > up so the users can share an Internet connection and a printer or two and > there are no servers at all. (You know, like you sometimes see in *school > computer labs*?) > ... so explain to me how NetworkManager.app fits into this picture. > It's not a "moronic" setup that people should be punished > for as Ronald claims; it's the real world, and generally the product of > economic decisions. If a school can either buy 30 iMacs, or 26 iMacs and a > G3 server, which do you think they'll pick? > 30 iMacs. Schools think like that, then generally regret it afterwards. At a school I'd pick 29 iMacs, and spend money on the 29th to add as much RAM and disk space to it, and run X-Server on it. > My point is that there are a lot of situations where users will not be > running a NetInfo network, and where you will not be able to convince them > to switch to a NetInfo network, and that Mac OS X has to not just > *tolerate* such situations, but *work well* with it. Heterogenous > networking is the reality, folks, get used to it. > We're already very used to it, thank you. Interoperability has been a watchword for a while. One of Jobs' throwaway lines from 6 years or so ago was "A while back everyone wanted computers on everyone's desk. Now they want just *one* computer on everyone's desk." mmalc. From zmonster at heliosgroup.com Wed Mar 3 16:08:50 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Why can't be included in a .M file? Message-ID: <5948F7268092D211B2020000D11C0FB808175C@chi-serv2.heliosgroup.com> In OpenStep NT 4.2, if I include in an Obj-C++ (.M) source file, the build fails miserably with a bunch of errors that say: parse error before '__stdcall' These errors pop up as all of the windows-related headers (i.e. windef.h, winbase.h, etc) get imported via the #include dependencies in those files. - Eric From Malcolm_Crawford at plsys.co.uk Wed Mar 3 16:19:12 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <199903032204.WAA03826@jasmine.plsys.co.uk> References: <199903032204.WAA03826@jasmine.plsys.co.uk> Message-ID: <199903040019.AAA03948@jasmine.plsys.co.uk> Oh well, not wanting to bring the debate back here, but to alert folks to a new set of pages under http://www.apple.com/macosx/server/ (*), I may have been talking about one thing, but it looks like Apple was delivering another: http://www.apple.com/macosx/server/afp.html Remote server administration Perhaps best of all, Apple File Services share the same administrative console. So wherever you happen to be, you can exercise complete control over your workgroup's network environment-remotely-from any web browser. mmalc. (*) Kudos to Joseph Shomphe of Skidmore College for spotting these first and mentioning them on the -talk list. From uli at haase-segel.de Thu Mar 4 00:13:29 1999 From: uli at haase-segel.de (Ulrich =?iso-8859-1?Q?K=F6ster?=) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X References: <9903031509.AA19805@california.uptime.ch> Message-ID: <36DE40AA.1A3063F2@haase-segel.de> *This message was transferred with a trial version of CommuniGate(tm) Pro* Philippe Robert schrieb: > *This message was transferred with a trial version of CommuniGate(tm) Pro* > You wrote: > > I think it's pretty obvious from the demo at MacWorld Tokyo that QuickTime > > streaming isn't quite...streaming just yet. > > > > OSXS ships w/ QT Stream Server (whatever name it'll have). > > But why don't they tell us this! So it's just an assumption... Digital Media Broadcasting Deliver movies to hundreds of Internet clients. Offer interactive teaching services to remote students. Provide more powerful training and customer service. The Mac OS X Server delivers the performance needed to support QuickTime HTTP streaming, compatible with all QuickTime 3 clients running on Macintosh and Windows platforms. You may ask where I got this information? http://www.apple.com/macosx. Ulrich From Philippe.Robert at uptime.ch Thu Mar 4 01:58:30 1999 From: Philippe.Robert at uptime.ch (Philippe Robert) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X In-Reply-To: <36DE40AA.1A3063F2@haase-segel.de> References: <36DE40AA.1A3063F2@haase-segel.de> Message-ID: <9903040958.AA00366@california.uptime.ch> You wrote: > > > OSXS ships w/ QT Stream Server (whatever name it'll have). > > But why don't they tell us this! So it's just an assumption... > > Digital Media > Broadcasting > Deliver movies to > hundreds of Internet > clients. Offer > interactive teaching > services to remote > students. Provide > more powerful > training and customer > service. The Mac OS X > Server delivers the > performance needed to > support QuickTime > HTTP streaming, > compatible with all > QuickTime 3 clients > running on Macintosh > and Windows > platforms. > > You may ask where I got this information? be included in a .M file? References: <5948F7268092D211B2020000D11C0FB808175C@chi-serv2.heliosgroup.com> Message-ID: <36DE6204.FFAE46E3@haase-segel.de> *This message was transferred with a trial version of CommuniGate(tm) Pro* Eric Hermanson schrieb: > *This message was transferred with a trial version of CommuniGate(tm) Pro* > In OpenStep NT 4.2, if I include in an Obj-C++ (.M) source file, > the build fails miserably with a bunch of errors that say: > > parse error before '__stdcall' > > These errors pop up as all of the windows-related headers (i.e. windef.h, > winbase.h, etc) get imported via the #include dependencies in those files. > > - Eric hi Eric I?m working with RDR2. And i remember right. This comes often when I have invisible keys in the file. Try to delete the whole line and the line before and enter your keys again. May be this helps Ulrich From uli at haase-segel.de Thu Mar 4 03:30:26 1999 From: uli at haase-segel.de (Ulrich =?iso-8859-1?Q?K=F6ster?=) Date: Thu Nov 3 14:41:52 2005 Subject: MacOS X Server v MacOS X References: <36DE40AA.1A3063F2@haase-segel.de> <9903040958.AA00366@california.uptime.ch> Message-ID: <36DE6ED2.26A8DA8F@haase-segel.de> *This message was transferred with a trial version of CommuniGate(tm) Pro* > phil wrote > > > Yes, I know, but this does *not* answer the question... I'd like to hear an > official statement from Apple that explains the reason for the delay! I?d like to hear an offical an offical statement from Apple that explains the reason for the delay, too. Who don`t? :-) As I remember Apple promised MacOS X (*not Server*) beta for Januar. There is also no word about it. Ulrich K?ster Heino Haase Segel GmbH Gneversdorfer Weg 9 23570 Travem?nde Germany Tel: +49 4502 2038 Fax: +49 4502 2315 http://www.haase-segel.de From uli at haase-segel.de Thu Mar 4 04:45:26 1999 From: uli at haase-segel.de (Ulrich =?iso-8859-1?Q?K=F6ster?=) Date: Thu Nov 3 14:41:52 2005 Subject: all problems solved Message-ID: <36DE8067.CD11ECE0@haase-segel.de> *This message was transferred with a trial version of CommuniGate(tm) Pro* My problems was: A router that dials up every time. An EOModeler saying a eomodeld file ist not a eomodeld file. A ProjectBuilder nover stoping indexing. And EOF Application don?t finding the eomodeld file in the framework And a DNS getting querys from Rhapsody like uli.haase.haase-segel.de 10.0.0.7.haase-segel.de 10.haase-segel.de. And believe me there are no hosts with that name. And there was the light .... I remebered that every DNS entry must have ending . I changed my hostname from uli.haase-segel.de to uli.haase-segel.de. . I?ve tested this for some minutes and no problems anymore. Ulrich From zmonster at heliosgroup.com Thu Mar 4 09:12:33 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Why won't this compile? Message-ID: <5948F7268092D211B2020000D11C0FB808175F@chi-serv2.heliosgroup.com> I've narrowed down the Objective-C++ compile problem I've been having in OpenStep 4.2 NT. This simple file won't compile when included in an 'app' type PB.project (similar files also won't work in a Framework project): File Name: test_main.M (note the capital 'M' indicating an Obj-C++ file) #import int main(int argc, const char *argv[]) {} I get numerous errors from all of the windows related headers (windef.h, winbase.h, etc). The errors are include: parse error before '__stdcall' ANSI C++ forbids declaration XXXX with no type or storage class I've tried numerous things, like setting/unsetting the -Obj-C++ CFLAG, wrapping #import in extern "C", extern "C++", extern "Objective-C" wrappers, etc. Nothing works. Any ideas? The above problem implies that one cannot compile Objective-C++ code that depends on any of the Windows headers (which obviously can't be the case or else C++ would be useless under OpenStep NT). Eric From respass at htl.com Thu Mar 4 09:37:05 1999 From: respass at htl.com (Marc Respass) Date: Thu Nov 3 14:41:52 2005 Subject: MXS partition size? Message-ID: <001701be6665$a0b04830$3290bad1@htl.com> Hi, I plan to get Mac OS X Server when it's available. I just got a new PowerBook with an eight GB hard drive. I'm partitioning it into two GB and six GB. I plan to use the two GB partition for MXS and the six GB partition for HFS+ and my Mac OS install. I'm thinking that since both operating systems can access the HFS+ partition, then I should be all set with two GB dedicated to MXS. What do you guys think? That also leads me to another question. Do you think (or know) that MXS will be able to take full advantage of HFS+ like long file names and security? Obviously, not much security since Mac OS will ignore it but longer file names would be cool. Thanks --Marc Respass From respass at htl.com Thu Mar 4 13:03:37 1999 From: respass at htl.com (Marc Respass) Date: Thu Nov 3 14:41:52 2005 Subject: MXS partition size Message-ID: <005401be6682$793dfdc0$3290bad1@htl.com> I wrote: > That also leads me to another question. Do you think (or know) that MXS will > be able to take full advantage of HFS+ like long file names and security? > Obviously, not much security since Mac OS will ignore it but longer file > names would be cool. Many people replied to me. Thanks for the replies. Seems that MXS will fully support HFS+ (woo-hoo) but, of course, Mac OS can read HFS+ and ignores all the additions to HFS+ so security is for naught and names longer than 31 characters are not supported. But that's OK. I can store documents and development project on the HFS+ partition. I can also move files between the two operating systems easily. Like a Downloads folder could be stored on the HFS+ partition. Thanks everyone -Marc Respass From respass at htl.com Thu Mar 4 13:06:27 1999 From: respass at htl.com (Marc Respass) Date: Thu Nov 3 14:41:52 2005 Subject: video drivers for MXS? Message-ID: <005e01be6682$e1655a10$3290bad1@htl.com> So I have this PowerBook and it rocks but the external video only goes to 1280x1024. I would love to have one of those ixMicro PC Card video adapters but I doubt it'll be supported under MXS. It also does not seem that ixMicro will be writing a driver (though they might). So is it difficult to write a video driver for MXS? Relate that to someone who has never written a driver and never built a standard Openstep application (me). I'm just thinking that it would be sweet to have this video card and maybe I can write the driver. It would be a good learning project. Thanks --Marc R From pierce at twinforces.com Thu Mar 4 13:22:49 1999 From: pierce at twinforces.com (Pierce T. Wetter III) Date: Thu Nov 3 14:41:52 2005 Subject: video drivers for MXS? In-Reply-To: <005e01be6682$e1655a10$3290bad1@htl.com> Message-ID: At 1:19 PM -0800 3/4/99, Marc Respass wrote: > So I have this PowerBook and it rocks but the external video only goes to > 1280x1024. I would love to have one of those ixMicro PC Card video adapters > but I doubt it'll be supported under MXS. It also does not seem that ixMicro > will be writing a driver (though they might). So is it difficult to write a > video driver for MXS? Relate that to someone who has never written a driver > and never built a standard Openstep application (me). I'm just thinking that > it would be sweet to have this video card and maybe I can write the driver. > It would be a good learning project. > > Thanks > --Marc R Check out ftp://ftp.apple.com/developer/macosxserver/docs/DriverKit.pdf and decide for yourself. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From dyoung at vviuh221.vvi.com Thu Mar 4 13:42:30 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:52 2005 Subject: video drivers for MXS? References: <005e01be6682$e1655a10$3290bad1@htl.com> Message-ID: <9903042142.AA01666@vviuh221.vvi.com> Marc, > So is it difficult to write a video driver for MXS? With good documentation any driver is a snap and a lot of fun. Well worth the midnight oil. With no documentation, as in what Apple proposes for Mac OS X Server, it is nearly impossible. List Members: If I have overlooked anything please correct me. I would also like to write drivers, but I don't see how it is possible. Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From paulrs at lgs-systems.com Thu Mar 4 12:26:17 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <9903042026.AA05787@slab> Hey guys, A little while back, someone else posted about big problems with a ballooning memory footprint for the WindowServer under YB NT. I just started running the YB NT that comes with the WOF 4.0 distribution and was very disturbed to find that not only is this true, but it's absurdly true. My window server, after having run only PB and TextEdit, is consuming over 40 MB of memory. Under 3.51, I don't recall ever seeing the window server get over 4-5 MB. I don't know how we can release our app if this isn't solved. I realize that the window server is disappearing in Mac OS X, but what can we do until then? Is this something that is being worked on, or has the price of our product just jumped by the cost of an extra 64 MB ram for each machine? Any thoughts would be welcome. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From epeyton at epicware.com Thu Mar 4 16:58:56 1999 From: epeyton at epicware.com (Eric S. Peyton) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine In-Reply-To: <9903042026.AA05787@slab> Message-ID: Paul, I think that your estimates have to be wrong. Look at them again. I run Yellow Box on NT WO 4.0 all day every day on a toshiba laptop with no performance degradation, etc. I am currently runnign TextEdit, Project Builder, EOModeler, And Stickies and have the following memory notes Application Name Mem Usage ProjectBuilder 9944 K EOModeler 9712 K Window Server 5596 K projectServer 4168 K Stickies 3704 K After that is all the window crap. I also currently have Photoshop and Explorer and MS Project and MS Word open on my machine. Granted I have 192 M RAM so that I can do this, adn this machine has not been rebooted since 5 PM yesterday. Eric On Thu, 4 Mar 1999, Paul R. Summermatter wrote: > Hey guys, > > A little while back, someone else posted about big problems with a > ballooning memory footprint for the WindowServer under YB NT. I just started > running the YB NT that comes with the WOF 4.0 distribution and was very > disturbed to find that not only is this true, but it's absurdly true. My > window server, after having run only PB and TextEdit, is consuming over 40 MB > of memory. Under 3.51, I don't recall ever seeing the window server get > over 4-5 MB. I don't know how we can release our app if this isn't solved. > I realize that the window server is disappearing in Mac OS X, but what can we > do until then? Is this something that is being worked on, or has the price > of our product just jumped by the cost of an extra 64 MB ram for each > machine? Any thoughts would be welcome. > > Regards, > Paul > > --- > > Paul Summermatter > > LGS Systems, Inc. > Medical Computing Division > > 15 TJ Gamester Ave > Portsmouth, NH 03801-5871 > (603) 433-9822 voice > (603) 433-9818 fax > (888) 898-6321 pager > 8986321@skytel.com paging email > > paulrs@lgs-systems.com > (NeXT or MIME Mail Welcome) > http://www.lgs-systems.com > > From zmonster at heliosgroup.com Thu Mar 4 18:43:56 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <5948F7268092D211B2020000D11C0FB8081767@chi-serv2.heliosgroup.com> I have _terrible_ performance problems with WebObjects 4.0 running on a Dell Inspiron. I was the one who originally commented on the out of control memory usage. If you keep PB open for > 12 hours, the WindowServer virtual memory usage gets out of control. It's routinely over 200 MB on my machine. I've noticed that running WOInfoCenter seems to be related to the memory problems - if this is not running, it's usually a little bit better, but not much. I "only" have 96MB of memory (with 300 MB swap space). I've noticed that people with >128MB RAM have few problems. I'm going to upgrade to 128 MB (maybe 196MB). The release notes for WO4.0 should indicate that 128MB+ of RAM is recommended. On a related issue, even in situations where the virtual memory usage of WindowServer+PB+WOInfoCenter is <100MB (I consider this a "good" day), my NT machine seems VERY sluggish when switching between PB windows (or any OpenStep windows for that matter). There definitely is a problem somewhere, and I don't think it's related to the setup of our machines (i.e. it's a problem in WO or WindowServer somewhere). Please note that I am not implying that Apple is doing a bad job. I am very impressed with WebObjects 4.0 and all the work everyone did to get it out (almost) on time. I just think a few performance tweaks are in order for the NT version. - Eric > -----Original Message----- > From: Eric S. Peyton [mailto:epeyton@epicware.com] > Sent: Thursday, March 04, 1999 7:05 PM > To: Multiple recipients of list > Subject: Re: Window Server Sucking the life out of my machine > > > Paul, > > I think that your estimates have to be wrong. Look at them > again. I run > Yellow Box on NT WO 4.0 all day every day on a toshiba laptop with no > performance degradation, etc. I am currently runnign > TextEdit, Project > Builder, EOModeler, And Stickies and have the following memory notes > > Application Name Mem Usage > ProjectBuilder 9944 K > EOModeler 9712 K > Window Server 5596 K > projectServer 4168 K > Stickies 3704 K > > After that is all the window crap. I also currently have > Photoshop and > Explorer and MS Project and MS Word open on my machine. > Granted I have > 192 M RAM so that I can do this, adn this machine has not > been rebooted > since 5 PM yesterday. > > Eric > > On Thu, 4 Mar 1999, Paul R. Summermatter wrote: > > > Hey guys, > > > > A little while back, someone else posted about big > problems with a > > ballooning memory footprint for the WindowServer under YB > NT. I just started > > running the YB NT that comes with the WOF 4.0 distribution > and was very > > disturbed to find that not only is this true, but it's > absurdly true. My > > window server, after having run only PB and TextEdit, is > consuming over 40 MB > > of memory. Under 3.51, I don't recall ever seeing the > window server get > > over 4-5 MB. I don't know how we can release our app if > this isn't solved. > > I realize that the window server is disappearing in Mac OS > X, but what can we > > do until then? Is this something that is being worked on, > or has the price > > of our product just jumped by the cost of an extra 64 MB > ram for each > > machine? Any thoughts would be welcome. > > > > Regards, > > Paul > > > > --- > > > > Paul Summermatter > > > > LGS Systems, Inc. > > Medical Computing Division > > > > 15 TJ Gamester Ave > > Portsmouth, NH 03801-5871 > > (603) 433-9822 voice > > (603) 433-9818 fax > > (888) 898-6321 pager > > 8986321@skytel.com paging email > > > > paulrs@lgs-systems.com > > (NeXT or MIME Mail Welcome) > > http://www.lgs-systems.com > > > > > From theisen at akaMail.com Fri Mar 5 01:37:24 1999 From: theisen at akaMail.com (Dirk Theisen) Date: Thu Nov 3 14:41:52 2005 Subject: video drivers for MXS? In-Reply-To: <005e01be6682$e1655a10$3290bad1@htl.com> Message-ID: <1do6ahu.1cu4xq8o5kruoM@ascend-tk-p25.rhrz.uni-bonn.de> Hi, Marc! I thought about buying a PowerBook as my main machine (as long as they have SCSI etc... :-)). > So I have this PowerBook and it rocks but the external video only goes to > 1280x1024. I got used to my two monitor setup and would like to preserve that. SO I also thought about buying the RoadRocket card (is there any other available at all?) > I would love to have one of those ixMicro PC Card video adapters > but I doubt it'll be supported under MXS. That's my problem as well. I'd like to switch to OXS as soon as possible... > It also does not seem that ixMicro > will be writing a driver (though they might). It might me worth asking... Regards, Dirk -- Buy a Pentium III now, get your personal Big Brother FREE!!! http://theisen.home.pages.de/ From jean-michel.cazaux at finindev.com Fri Mar 5 03:11:05 1999 From: jean-michel.cazaux at finindev.com (Jean-Michel Cazaux) Date: Thu Nov 3 14:41:52 2005 Subject: EOPopupAssociation Problem Message-ID: <01BE6701.5D591520@JEAN-MICHEL> Hi everybody, To edit a to-one relationship, I use a pop-up and a EOPopUpAssociation. I connect the "titles" aspect of the association to the property where titles can be found in the displayGroup where objects for the destination of the relationship are fetched ; I connect the "selectedObject" aspect on the "toMyRelation" property of the displayGroup where the objects I want to edit are. This seems correct and it had worked fine in other cases... But here, when I load the nib (fetch the displayGroups), I got the following exception : Mar 05 11:33:14 AnalysTH[307] *** Assertion failure in -[NSMenu insertItemWithTitle:action:keyEquivalent:atIndex:], NSMenu.m:378 Mar 05 11:33:14 AnalysTH[307] *** NSRunLoop: ignoring exception 'NSInternalInconsistencyException' (reason 'Invalid parameter not satisfying: (index >= 0) && (index <= [_itemArray count])') that raised during delayed perform of target 17880288 and selector '_processEndOfEventObservers:' The stranger thing is that it works if I got only a few objects (less than 6/7) in the 'titles' displayGroup.... Another point : for this purpose, I use the flatFile adaptor (don't think it makes a difference in this case)... Thanks in adavance for a quick answer, I'm getting crazy !!!!! :-( __________________________________________________________________________________________ Jean-Michel CAZAUX FININDEV, Conseil en Finances pour les Collectivit?s Locales. 204 Rue Michel Teule - ZAC d'Alco 34080 Montpellier - FRANCE. T?l. +33 (0)4 67 63 66 25 - Fax +33 (0)4 67 63 35 45 e-mail: jean-michel.cazaux@finindev.com From paulrs at lgs-systems.com Fri Mar 5 03:44:15 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine In-Reply-To: References: Message-ID: <9903051144.AA06305@slab> Eric Peyton wrote: [ Paul, [ [ I think that your estimates have to be wrong. Look at them again. Eric et al, They are most definitely not wrong!!! However, I did find that the window server memory usage was significantly impacted by the number of projects open. Because I was not compiling our app from the command line, I had about 10-15 projects open at once. I found that as soon as I quit project builder, the window server's memory usage went right back down to where I would expect it to be. Luckily, our app can be run from a base of three windows, so maybe that is why we've never really encountered this issue. Regards, Paul [ I run [ Yellow Box on NT WO 4.0 all day every day on a toshiba laptop with no [ performance degradation, etc. I am currently runnign TextEdit, Project [ Builder, EOModeler, And Stickies and have the following memory notes [ [ Application Name Mem Usage [ ProjectBuilder 9944 K [ EOModeler 9712 K [ Window Server 5596 K [ projectServer 4168 K [ Stickies 3704 K [ [ After that is all the window crap. I also currently have Photoshop and [ Explorer and MS Project and MS Word open on my machine. Granted I have [ 192 M RAM so that I can do this, adn this machine has not been rebooted [ since 5 PM yesterday. [ [ Eric [ [ On Thu, 4 Mar 1999, Paul R. Summermatter wrote: [ [ > Hey guys, [ > [ > A little while back, someone else posted about big problems with a [ > ballooning memory footprint for the WindowServer under YB NT. I just [ > started running the YB NT that comes with the WOF 4.0 distribution and [ > was very disturbed to find that not only is this true, but it's [ > absurdly true. My window server, after having run only PB and [ > TextEdit, is consuming over 40 MB of memory. Under 3.51, I don't [ > recall ever seeing the window server get over 4-5 MB. I don't know how [ > we can release our app if this isn't solved. I realize that the window [ > server is disappearing in Mac OS X, but what can we do until then? Is [ > this something that is being worked on, or has the price of our [ > product just jumped by the cost of an extra 64 MB ram for each [ > machine? Any thoughts would be welcome. [ > [ > Regards, [ > Paul [ > [ > --- [ > [ > Paul Summermatter [ > [ > LGS Systems, Inc. [ > Medical Computing Division [ > [ > 15 TJ Gamester Ave [ > Portsmouth, NH 03801-5871 [ > (603) 433-9822 voice [ > (603) 433-9818 fax [ > (888) 898-6321 pager [ > 8986321@skytel.com paging email [ > [ > paulrs@lgs-systems.com [ > (NeXT or MIME Mail Welcome) [ > http://www.lgs-systems.com [ > [ > [ --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From pierce at alumni.caltech.edu Fri Mar 5 07:58:26 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine In-Reply-To: <9903051144.AA06305@slab> Message-ID: At 5:10 AM -0800 3/5/99, Paul R. Summermatter wrote: > Eric Peyton wrote: > [ Paul, > [ > [ I think that your estimates have to be wrong. Look at them again. > > Eric et al, > > They are most definitely not wrong!!! However, I did find that the > window server memory usage was significantly impacted by the number of > projects open. Because I was not compiling our app from the command line, I > had about 10-15 projects open at once. I found that as soon as I quit > project builder, the window server's memory usage went right back down to > where I would expect it to be. > > Luckily, our app can be run from a base of three windows, so maybe > that is why we've never really encountered this issue. Perhaps its the backing store save! If you're on a 24-bit color system, and you have 10-15 projects open, that's the memory right there. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From john.zollinger at arkona.com Fri Mar 5 10:25:16 1999 From: john.zollinger at arkona.com (John Zollinger) Date: Thu Nov 3 14:41:52 2005 Subject: Jikes java compiler in PB Message-ID: <000401be6735$84a18ca0$1600a8c0@katana.arkona.com> Has anyone written a compiler profile (for use with PB's buildfilter) that understands the output from IBM's Jikes Java compiler? Not being a regex wizard I'm having troubles getting it to work properly. :-/ Is anyone else using PB as their Java IDE? I think it is far better than the majority of the tools out there in that role. I wish Apple would sell it outside the realm of WOF, I think a lot of people would really like it. It's a great tool. Thanks! John Zollinger V.P. Engineering Arkona, Inc. john.zollinger@arkona.com "Enabling the Extended Enterprise" http://www.arkona.com/ From dave.pl at ping.at Fri Mar 5 09:56:51 1999 From: dave.pl at ping.at (Dietmar Planitzer) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903051851.TAA26336@pong.Austria.EU.net> >From: "Pierce T. Wetter III" > >At 5:10 AM -0800 3/5/99, Paul R. Summermatter wrote: >> Eric Peyton wrote: >> [ Paul, >> [ >> [ I think that your estimates have to be wrong. Look at them again. >> >> Eric et al, >> >> They are most definitely not wrong!!! However, I did find that the >> window server memory usage was significantly impacted by the number of >> projects open. Because I was not compiling our app from the command >line, I >> had about 10-15 projects open at once. I found that as soon as I quit >> project builder, the window server's memory usage went right back down to >> where I would expect it to be. >> >> Luckily, our app can be run from a base of three windows, so maybe >> that is why we've never really encountered this issue. > > Perhaps its the backing store save! If you're on a 24-bit color >system, and you have 10-15 projects open, that's the memory right >there. > That is most likely the problem. On my RDR2 for PPC the WindowServer has 43 bitmaps (used for backing store) allocated even after a fresh reboot ! If you want to have an exact statistics about the amount of memory the WindowServer has currently allocated for the various bitmaps, framebuffer, execution contexts private and global VM then simply execute the "vmusage" operator in the pft tool. Its interesting to note that by far the vast majority of the RAM used by the WindowServer is used for backing store. The PS engine itself occupies only around 10% - 20% of that memory. Regards, Planitzer Dietmar From sebestyen_g at accentcomm.com Fri Mar 5 10:51:33 1999 From: sebestyen_g at accentcomm.com (Gabor Sebestyen) Date: Thu Nov 3 14:41:52 2005 Subject: PB/CVS - how to start? Message-ID: <36E027B5.B360DE45@accentcomm.com> Hi! I'm learning CVS version management and I want to understand how make PB to cooperate with CVS. Just a novice question: How can I add a new project to the repository? And ... What is the first step to put a new project into control of CVS? Put it manually using 'cvs import ...' and try to get into PB from the repository? Or open my project in PB and put it into the repository using CVS SCM bundle. Finally ... Docs? Where can I find (if exists) one? Thank you! Best regards, Gabor -------------------------------------------------------------- A C C E N T C O M M U N I C A T I O N S - We Make Your Web Groove - Tel: (+36 1) 315-0701 http://www.accentcomm.com (+36 1) 343-4440 e-mail: info@accentcomm.com GSM: (+36 30) 951-4334 Fax: (+36 30) 980-0344 -------------------------------------------------------------- From bbum at codefab.com Fri Mar 5 11:08:00 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine References: <199903051851.TAA26336@pong.Austria.EU.net> Message-ID: <36E02B8F.B81A417B@mail.codefab.com> The WindowServer does suck resources left and right on NT-- just about any app does (Netscape certainly chews memory). Upgrade to 256mb and the machine should never swap again (though mine mysteriously bumps to 400MB of vm every afternoon at 4:00pm-- unrelated to OpenStep, however). The problem is three-fold: (1) NT's virtual memory management system completely sucks. It is one of the least effecient algorithms around and it has a tendency to swap out entire processes, if it can. Under Mach, if you overrun the system memory by 20 to 30 percent-- even 50 percent-- performance is still acceptable. Not under NT; as soon as it starts swapping, you're system is going to suck. (if you want a detailed explanation of why the VM on NT sucks, I can have Chuck Swiger post it-- he explained it to me, but I wouldn't trust myself to reexplain it properly) (2) NT's intra-process communication model sucks, as well. Especially when two processes have to move lots of memory back and forth (or even just forth). Unfortunately, the WindowServer model used throughout the current OpenStep totally relies on effecient intra-application communication. Under mach, when the app needs to hand off a bitmap to the windowserver, it simply sends a mach message with the address of the bitmap-- this memory is mapped into the windowserver WITHOUT COPYING IT by using the Copy-On-Write features of Mach's memory manager. Under NT, the bitmap is effectively copied. [since the WindowServer is Going Away in the near future, this problem should Go Away, as well] (3) The default configuration of Windows NT is to give the forground process maximum priority when scheduling CPU cycles. Fine if your app doesn't constantly talk to some background process all the time. Of course, every UI app under OpenStep has to talk to the windowserver all the time to get drawing done. If you right-click on My Computer then click the Performance tab, you can set the scheduling of the foreground app to "normal performance"-- this will greatly improve the situation. OpenStep was designed around a kernel that was quite effecient at memory management and intra-process communication. NT does neither effeciently and very few third party developers in the WIN32 marketplace write applications that stress either feature. Apple's announced direction for OpenStep-- though unfortunate in some respects (a true window manager is a very useful beast, indeed)-- will definitely address these problems. b.bum Dietmar Planitzer wrote: > >From: "Pierce T. Wetter III" > > > >At 5:10 AM -0800 3/5/99, Paul R. Summermatter wrote: > >> Eric Peyton wrote: > >> [ Paul, > >> [ > >> [ I think that your estimates have to be wrong. Look at them again. > >> > >> Eric et al, > >> > >> They are most definitely not wrong!!! However, I did find that the > >> window server memory usage was significantly impacted by the number of > >> projects open. Because I was not compiling our app from the command > >line, I > >> had about 10-15 projects open at once. I found that as soon as I quit > >> project builder, the window server's memory usage went right back down to > >> where I would expect it to be. > >> > >> Luckily, our app can be run from a base of three windows, so maybe > >> that is why we've never really encountered this issue. > > > > Perhaps its the backing store save! If you're on a 24-bit color > >system, and you have 10-15 projects open, that's the memory right > >there. > > > > That is most likely the problem. On my RDR2 for PPC the WindowServer > has 43 bitmaps (used for backing store) allocated even after a fresh > reboot ! > > If you want to have an exact statistics about the amount of memory the > WindowServer has currently allocated for the various bitmaps, > framebuffer, execution contexts private and global VM then simply > execute the "vmusage" operator in the pft tool. > > Its interesting to note that by far the vast majority of the RAM used > by the WindowServer is used for backing store. The PS engine itself > occupies only around 10% - 20% of that memory. > > Regards, > > Planitzer Dietmar From chuck at codefab.com Fri Mar 5 12:57:30 1999 From: chuck at codefab.com (Chuck Swiger) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903052057.PAA03522@tertius.codefab.com> As bbum got several requests, I dug up some Usenet articles on why Windows NT doesn't utilize memory very well. See: http://www.dejanews.com/[ST_rn=ps]/getdoc.xp?AN=321819777 To provide a quick summary, the problem is that NT uses a FIFO page selection algorithm with working set modelling. FIFO is one of the worst choices but the addition of the working set notion gives NT mediocre (barely) VM performance. FIFO suffers from Belady's anomaly, which is the counterintuitive behavior that making more memory available to a process may actually make the process swap more. By contrast, Unix systems use the clock hand algorithm, which is a stack-based NFU (not frequently used) algorithm which does not suffer from Belady's anomaly. That means the system can better use a global algorithm to allocate memory _between_ all running processes...which is something NT is notably bad at. -Chuck Charles Swiger | chuck@codefab.com | Yeah, yeah-- disclaim away. ----------------+-------------------+---------------------------- You have come to the end of your journey. Survival is everything. From louis.pelletier at altavista.net Fri Mar 5 13:36:51 1999 From: louis.pelletier at altavista.net (Louis Pelletier) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903052135.QAA21225@wacky.total.net> Thanks for these explanations! :o) Now I understand why when my Pentium 400 with UW SCSI running Windows NT Server can't move my mouse in real-time when the OS start swapping (while I have 128 MB or RAM)... Suddently, the unexplainable becomes logical! lol Isn't Microsoft products FUN!!!! 8-) -- Louis Pelletier, Qu?bec ---------- >De?: "Chuck Swiger" >? : Multiple recipients of list >Objet?: Re: Window Server Sucking the life out of my machine >Date?: Vend 05 mars 1999 16:08 > > As bbum got several requests, I dug up some Usenet articles on why Windows > NT doesn't utilize memory very well. See: > > http://www.dejanews.com/[ST_rn=ps]/getdoc.xp?AN=321819777 > > To provide a quick summary, the problem is that NT uses a FIFO page > selection algorithm with working set modelling. FIFO is one of the worst > choices but the addition of the working set notion gives NT mediocre (barely) > VM performance. FIFO suffers from Belady's anomaly, which is the > counterintuitive behavior that making more memory available to a process may > actually make the process swap more. > > By contrast, Unix systems use the clock hand algorithm, which is a > stack-based NFU (not frequently used) algorithm which does not suffer from > Belady's anomaly. That means the system can better use a global algorithm to > allocate memory _between_ all running processes...which is something NT is > notably bad at. > > -Chuck > > Charles Swiger | chuck@codefab.com | Yeah, yeah-- disclaim away. > ----------------+-------------------+---------------------------- > You have come to the end of your journey. Survival is everything. > > From btoback at optc.com Fri Mar 5 14:30:39 1999 From: btoback at optc.com (Bruce Toback) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903052232.PAA11235@landru.optc.com> Louis Pelletier writes: >Now I understand why when my Pentium 400 with UW SCSI running Windows NT >Server can't move my mouse in real-time when the OS start swapping (while I >have 128 MB or RAM)... > >Suddently, the unexplainable becomes logical! lol > >Isn't Microsoft products FUN!!!! 8-) No. [Long, loud rant deleted since this is the -dev list] My solution to the problem of being unable to compile and edit at the same time on my 128MB Pentium-166 was to buy a dual-processor 350MHz machine for $2,000 or so. Now I can compile and edit at the same time, just like I can on my $65 Linux box. (Yes, $65.) Or, for that matter, on my 10-year-old Mac IIx, which doesn't even *pretend* to have preemptive multitasking. -- Bruce -------------------------------------------------------------------------- Bruce Toback Tel: (602) 996-8601| My candle burns at both ends; OPT, Inc. (800) 858-4507| It will not last the night; 11801 N. Tatum Blvd. Ste. 142 | But ah, my foes, and oh, my friends - Phoenix AZ 85028 | It gives a lovely light. btoback@optc.com | -- Edna St. Vincent Millay From zmonster at heliosgroup.com Fri Mar 5 14:43:58 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <5948F7268092D211B2020000D11C0FB808176E@chi-serv2.heliosgroup.com> > -----Original Message----- > From: Bruce Toback [mailto:btoback@optc.com] >> Louis Pelletier writes: > >Isn't Microsoft products FUN!!!! 8-) > > No. > > [Long, loud rant deleted since this is the -dev list] > > My solution to the problem of being unable to compile and edit at the > same time on my 128MB Pentium-166 was to buy a dual-processor 350MHz > machine for $2,000 or so. This thread has made me even more (excuse the French) _f-ing_ sick of NT. I cannot believe that any organization in their right mind would choose to use NT as a server, let alone a workstation OS. Ignoring the fact that NT's memory model totally sucks and makes it completely useless when doing "production-level" things, the mere fact that NT lacks the following things inherently de-classifies it as a server OS: - no symbolic linking (forget about sharing files - must COPY shared resources every-f-ing-where) - boneheaded NTFS (CANNOT install a new version of a shared resource unless ALL applications using that resource are killed). This puts the sysadmin in the position of EITHER copying the resource on ALL user machines across the network, or, asking all users (whether it's 1, 10, 100, or 1,000,000) to kill off their applications so a new dll can be installed). On UNIX, you simply install the new resource, and existing apps use the old resource while newly started apps use the new resource (with the same name - the UNIX filesystem stores a new 'pointer' to the updated resource). - one must REBOOT the NT machine whenever a new software product is installed!! Gee that's REALLY condusive to easy system administration!! - another "nice" surprise is that NT COPIES your registry across the network EVERY time you log in AND log out. If you're registry is as big as some, you've got about 80 MB of copying to wait for. That list just BEGINS to outline how bad NT sucks. Please Apple, position MacOSX dead-straight against NT and bury that P.O.S.! Eric From louis.pelletier at altavista.net Fri Mar 5 15:15:15 1999 From: louis.pelletier at altavista.net (Louis Pelletier) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903052314.SAA15763@pablo.total.net> Obviously, this thread as nothing to do on this list, but since I find your last implication interesting, I've moved it to the "talk" list. -- Louis Pelletier, Qu?bec ---------- >De?: Eric Hermanson >? : Multiple recipients of list >Objet?: RE: Re : Window Server Sucking the life out of my machine >Date?: Vend 05 mars 1999 17:59 > > >> -----Original Message----- >> From: Bruce Toback [mailto:btoback@optc.com] > >>> Louis Pelletier writes: > >> >Isn't Microsoft products FUN!!!! 8-) >> >> No. >> >> [Long, loud rant deleted since this is the -dev list] >> >> My solution to the problem of being unable to compile and edit at the >> same time on my 128MB Pentium-166 was to buy a dual-processor 350MHz >> machine for $2,000 or so. > > This thread has made me even more (excuse the French) _f-ing_ sick of NT. I > cannot believe that any organization in their right mind would choose to use > NT as a server, let alone a workstation OS. Ignoring the fact that NT's > memory model totally sucks and makes it completely useless when doing > "production-level" things, the mere fact that NT lacks the following things > inherently de-classifies it as a server OS: > > - no symbolic linking (forget about sharing files - must COPY shared > resources every-f-ing-where) > > - boneheaded NTFS (CANNOT install a new version of a shared resource > unless ALL applications using that resource are killed). This puts the > sysadmin in the position of EITHER copying the resource on ALL user machines > across the network, or, asking all users (whether it's 1, 10, 100, or > 1,000,000) to kill off their applications so a new dll can be installed). > On UNIX, you simply install the new resource, and existing apps use the old > resource while newly started apps use the new resource (with the same name - > the UNIX filesystem stores a new 'pointer' to the updated resource). > > - one must REBOOT the NT machine whenever a new software product is > installed!! Gee that's REALLY condusive to easy system administration!! > > - another "nice" surprise is that NT COPIES your registry across the > network EVERY time you log in AND log out. If you're registry is as big as > some, you've got about 80 MB of copying to wait for. > > > That list just BEGINS to outline how bad NT sucks. Please Apple, position > MacOSX dead-straight against NT and bury that P.O.S.! > > Eric > > From ajetha at tacticalstep.com Fri Mar 5 16:02:28 1999 From: ajetha at tacticalstep.com (A Jetha) Date: Thu Nov 3 14:41:52 2005 Subject: Calendar view Message-ID: <9903051902.AA2811810@avatar> Folks, Just wondering if there is a reusable calendar view on the net somewhere? I remember there was on in the NeXTSTEP misckit, but it's not to be found in the OpenStep misckit. If someone knows of a OpenStep compliant version, could you let me know please. Thanks -- Alykhan Jetha (AJ) TacticalStep Inc., Email: mailto:ajetha@tacticalstep.com WWW: http://www.tacticalstep.com Tel: 905-831-2643 Fax: 905-837-2546 From discord at charm.net Fri Mar 5 18:03:39 1999 From: discord at charm.net (dIsCoRd) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine In-Reply-To: <199903052057.PAA03522@tertius.codefab.com> Message-ID: On Fri, 5 Mar 1999, Chuck Swiger wrote: > To provide a quick summary, the problem is that NT uses a FIFO page > selection algorithm with working set modelling. FIFO is one of the worst > By contrast, Unix systems use the clock hand algorithm, which is a > stack-based NFU (not frequently used) algorithm which does not suffer from Just as followup, the only reference that I have (OS COncepts, 4th+5th, Silberschatz and Galvin - the dinosaur book :) states that NT uses per-process FIFO with WS modeling and a default WS of 30 pages. This is tested occaisionally by NT "stealing" pages and testing if the process faults for that page. Mach (up to v3), uses FIFO with 2nd chance with WS modeling. No note of the default WS. It does note that Mach support user level page management, so it is possible that BSD on top of Mach is using something else entirely. -- /\ karl hsu / \ discord@charm.net / \ www.charm.net/~discord/ / () \ / \ / \____/ \ /____________\ From louis.pelletier at altavista.net Fri Mar 5 19:25:01 1999 From: louis.pelletier at altavista.net (Louis Pelletier) Date: Thu Nov 3 14:41:52 2005 Subject: Window Server Sucking the life out of my machine Message-ID: <199903060323.WAA22843@pablo.total.net> Anyway, does all those concern applies to MacOS X? As I've understand it, Apple is working on Mach to make it better. And since Tevanian is the head of both Mach and MacOS, I don't think OS X will come out with crappy VM implementation ? la Windows NT... -- Louis Pelletier, Qu?bec ---------- >De?: dIsCoRd >? : Multiple recipients of list >Objet?: Re: Window Server Sucking the life out of my machine >Date?: Vend 05 mars 1999 21:14 > > On Fri, 5 Mar 1999, Chuck Swiger wrote: > >> To provide a quick summary, the problem is that NT uses a FIFO page >> selection algorithm with working set modelling. FIFO is one of the worst > >> By contrast, Unix systems use the clock hand algorithm, which is a >> stack-based NFU (not frequently used) algorithm which does not suffer from > > > Just as followup, the only reference that I have (OS COncepts, 4th+5th, > Silberschatz and Galvin - the dinosaur book :) states that > > NT uses per-process FIFO with WS modeling and a default WS of 30 pages. > This is tested occaisionally by NT "stealing" pages and testing if the > process faults for that page. > > Mach (up to v3), uses FIFO with 2nd chance with WS modeling. No note of > the default WS. It does note that Mach support user level page management, > so it is possible that BSD on top of Mach is using something else > entirely. > > -- > > /\ karl hsu > / \ discord@charm.net > / \ www.charm.net/~discord/ > / () \ > / \ > / \____/ \ > /____________\ > > > From paulrs at lgs-systems.com Sat Mar 6 16:09:26 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:52 2005 Subject: No visual cue that pop-up menu is first responder Message-ID: <9903070009.AA07446@slab> RE: YB NT / WOF 4.0 Hey guys, Is there a good reason why I do not see a dotted outline around pop-up buttons & pull down menus when they are the first responder? I hope this is a bug which is being addressed and not a new interface guidline. It seems odd that regular buttons still show a dotted outline, but pop ups don't. Is this going to be my first poseAs, or should we just wait for the next update? Thanks, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From kelleysoft at pressenter.com Sun Mar 7 06:18:35 1999 From: kelleysoft at pressenter.com (Greg Betzel) Date: Thu Nov 3 14:41:52 2005 Subject: archiving a window In-Reply-To: <9903070009.AA07446@slab> Message-ID: Hi, I have an NSWindow, created manually (not by nib) that gets a good number of views & controls added to it over time. My question is, is it appropriate to archive the whole works by just calling [theArchiver encodeObject:theWindow], and still have everything decode correctly? Assume that encodeRootObject: has already been called, and the window is getting archived in the middle of a graph. I ask because my app is having some strange problems with NSViews archived/unarchived this way, that it doesn't have with views created manually or loaded from a nib. For example, setFrameSize: on a good view was bringing the whole app down. Talk about fustrating.... Thanks in advance. Greg Betzel From dyoung at vviuh221.vvi.com Sun Mar 7 06:42:50 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:52 2005 Subject: archiving a window References: Message-ID: <9903071442.AA04460@vviuh221.vvi.com> Greg, > My question is, is it appropriate to archive the whole works by > just calling [theArchiver encodeObject:theWindow] The answer is ... yes and no ... yes, it should work for the version you are working with. no, it may not work with other versions. This issue has a long history of yes, no answers some of which are: - Next said don't do it. - Next said you can do it. - Next had examples of code where it was explicitly commented not to do it. - Transition of window archiving from Nextstep 2 to 3 broke some people's document archiving. - It works [now] so why not use it. - Every object should archive, including a window. - When you open the document archive (using a text editor) of many company's document archive you find they have encoded the window object into it. - We elected not to archive the window object into our documents because it was too risky. There was a long debate of whether to do it or not. My 2 cents: I would steer clear of archiving instances of the NSWindow class for shrinkwrap, but for proof of concept it should be fine (last I checked). Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From kelleysoft at pressenter.com Sun Mar 7 09:26:35 1999 From: kelleysoft at pressenter.com (Greg Betzel) Date: Thu Nov 3 14:41:52 2005 Subject: archiving a window In-Reply-To: <9903071442.AA04460@vviuh221.vvi.com> Message-ID: David, Hmm. Interesting. >- Every object should archive, including a window. That's what I was assuming... >There was a long debate of whether to do it or not. My 2 cents: I would >steer clear of archiving instances of the NSWindow class for shrinkwrap, but >for proof of concept it should be fine (last I checked). This is destined for shrinkwrap, so I'll take your advice (especially since I'm getting strange and intermittent problems). This has to work solidly & reliably "here on out"...don't need to be taking any chances with "well, it happens to work now" code. :) Thanks. Greg Betzel From paulrs at lgs-systems.com Sun Mar 7 10:05:49 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:52 2005 Subject: Outline View in IB Message-ID: <9903071805.AA07808@slab> RE: YB NT, WOF 4.0 Hey guys, Does anyone know why the outline view for the instances in IB no longer displays the name of the object in parenthesis (as it used to)? This is creating a big problem for me. Is there a preference for this? Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From paulrs at lgs-systems.com Sun Mar 7 12:16:39 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:52 2005 Subject: Submitting bugs Message-ID: <9903072016.AA07891@slab> Hey guys, Where are we supposed to submit bugs regarding YB NT? Is there any breakdown regarding product line, WOF, EOF, AppKit, Foundation, PDO, etc? Thanks, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From pbjones at bigpond.com Sun Mar 7 18:34:57 1999 From: pbjones at bigpond.com (PBJ) Date: Thu Nov 3 14:41:52 2005 Subject: choices NT/RDR2/MXS Message-ID: A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 678 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990308/f2f126a0/attachment.bin From rcfa at cubiculum.com Sun Mar 7 18:52:09 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:52 2005 Subject: choices NT/RDR2/MXS In-Reply-To: References: Message-ID: <199903080252.AA01882@kannix.cubiculum.com> > Will people still be developing for RDR2 and OpenStep once MXS ships? RDR2: hardly OpenStep: possibly > Is YB on NT seen as the Intel replacement of OpenStep/RDR2-Intel? Not by any sane person, unless forced by red hot coals (or pointy haired manager types) > What do people see as an upgrade for OpenStep? ('Nothing' can be a valid > answer) At least for the intel version nothing is the answer. GNUStep isn't functional enough yet to be considered either upgrade or just replacement. OS/YB-WinNT is a definite downgrade, sticking with what we have is a no-grade, and OSXS isn't an option, because Apple won't sell it to us. As for OSXS-PPC, it's a new product on a new platform, and thus neither up-, side-, or downgrade. It's simply new. (Anything that requires both new hard and software is new in that sense, because it orphans previous markets.) > I would like to develop for OS/MSX/RDR2/YB but I have to make a big Hardware > commitment first. (ie either G3 or PII) At this point, if you're not already committed to the platform, I'd wait. Intel is dead as far as Apple goes. So if you go PII/PIII, you might be able to work on GNUStep, which is fine if it's not either involving GUI stuff or production release software. If you chose NT, you chose your own punishment. If you go with Apple, you should prepare yourself for abuse by a company that has a history of breaking its word. Take your pick, none seems very attractive to me at this point... GNUStep wins in so far, as that it's based entirely on free, open-source software, and at least if something is missing, you can add it, and you know nothing you have will be ever taken away... Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From Barry.Leslie at snap.de Mon Mar 8 00:25:32 1999 From: Barry.Leslie at snap.de (Barry Leslie) Date: Thu Nov 3 14:41:52 2005 Subject: Problem linking framwork. Message-ID: <36E38978.115C@PrimeBase.com> The linker on Mac OS X is linking my framework incorrectly. I also had this problem with DR 2. It worked on DR1. What happens is that when function Foo() should be called some other function is called, or if I rearrange the link order a global variable will be called as if it where a function. I have tried renaming things but it has no effect. I can step through the code in the debugger and everything is OK up until I call Foo() and I suddenly find myself some where else in the code. This code works fine when it is linked into an application. Is this a linker bug or is there some linker setting that I am missing. Barry From frank at inis.de Mon Mar 8 05:02:48 1999 From: frank at inis.de (Frank Knobloch) Date: Thu Nov 3 14:41:52 2005 Subject: What does the following warning mean? Message-ID: <199903081403.PAA01335@inis-9.inis.de> > warning: > duplicate interface declaration for class `ISSystemControlledProcess' It's a small class in a subproject. How can i get such a warning? Thanks Frank From janos.lobb at yale.edu Mon Mar 8 06:45:51 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:52 2005 Subject: Fwd: Problem linking framwork. Message-ID: >Date: Mon, 8 Mar 1999 00:41:46 -0800 (PST) >Reply-To: Barry.Leslie@snap.de >Originator: macosx-dev@omnigroup.com >Sender: macosx-dev@omnigroup.com >Precedence: bulk >From: "Barry Leslie" >To: Multiple recipients of list >Subject: Problem linking framwork. >X-Comment: To unsubscribe, follow directions at >http://www.omnigroup.com/MailArchive/ >MIME-Version: 1.0 >Status: > >The linker on Mac OS X is linking my framework incorrectly. I also had >this problem with DR 2. It worked on DR1. I have just DR2. How, where and when did you get Mac OS X ? > >What happens is that when function Foo() should be called some other >function is called, or if I rearrange the link order a global variable >will be called as if it where a function. I have tried renaming things >but it has no effect. > >I can step through the code in the debugger and everything is OK up >until I call Foo() and I suddenly find myself some where else in the >code. This code works fine when it is linked into an application. > >Is this a linker bug or is there some linker setting that I am missing. > >Barry > J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From gerard at ina.fr Mon Mar 8 12:46:05 1999 From: gerard at ina.fr (Gerard Iglesias) Date: Thu Nov 3 14:41:52 2005 Subject: Stopping the default "new" document in Document-Based apps References: <199902271837.KAA30373@boom.lorax.com> Message-ID: <36E4370C.90967044@ina.fr> Hi, I am coming back from holidays, hence I can, only now, react to this post. mike wrote: > Gerard wrote > > For example we have a multi doc app which contains a lot of inspectors > > (View/Controller) that need to be synchronized with the current > main doc > > and I had a lot of difficulty to have the thing together (it was the > > principal reason for what I subclassed NSDocumentController). In fact > > the currentDocument method can return nil, because it looks at the main > > window which can be nil in certain circumstances. > > > > I recommend looking at the Draw2/Sketch example (Draw2 in DR2, > Sketch in OSXS). > I looked at today, and I have to say that it is not the better way to do the V/C synchronization, from my humble point of view. It seems to be a hack. > In general an NSDocument should own the NSWindowControllers for all > the windows that belong exclusively to the document. > NSWindowControllers for ancillary panels like inspectors should not > be owned by a document. OK. > > Draw2/Sketch shows how to use NSDocuments and NSWindowControllers in > this way. DrawDocument creates and owns DrawWindowControllers which > manage the main document window for each document. The > InspectorController class is not created/owned by any document. It > implements a sharedInstance API that allows the one > InspectorController to be found and it observes what is going on in > the app to keep its state up to date and in sync with whatever > document is currently active. There are many ways that an inspector > can figure out what to inspect. > Maybe, but what I would want, is a common and simple way to do that, I want to be notified by the document that it just did become the main document. What is made in the Sketch inspector is to be registered of the NSWindowDidBecomeMainNotification notification, and when the notifcation is received it test the class of the windowController of the window to decide if it needs to change is fucus. I don't think that it is a good example of object design, it is almost a C hack :-) > I don't believe it should be necessary to subclass > NSDocumentController in order to implement inspectors reasonably. It is my way, and regarding your arguments, I am now convinced that it is the good way. My NSDocumentcontroller subclass communicate with the documents to manage a clean way to synchronize the registered View/Controller with the model. In my subclassing I added to the subclass the reponsability to manage the differents bundles which are View/Controller of the main document. > There is nothing wrong with subclassing NSDocumentController > (although it is perhaps harder than it ought to be currently to get > your subclass to be the one that creates the shared instance.) I don't understand that, I copied the Sketch way: instantiate it in the app main nib file. Sincerely. Gerard. From sanguish at digifix.com Mon Mar 8 14:21:33 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:52 2005 Subject: Stopping the default "new" document in Document-Based apps In-Reply-To: <36E4370C.90967044@ina.fr> References: <36E4370C.90967044@ina.fr> Message-ID: <199903082221.RAA26926@digifix.com> Gerard Iglesias wrote: > > > > Draw2/Sketch shows how to use NSDocuments and NSWindowControllers in > > this way. DrawDocument creates and owns DrawWindowControllers which > > manage the main document window for each document. The > > InspectorController class is not created/owned by any document. It > > implements a sharedInstance API that allows the one > > InspectorController to be found and it observes what is going on in > > the app to keep its state up to date and in sync with whatever > > document is currently active. There are many ways that an inspector > > can figure out what to inspect. > > > > Maybe, but what I would want, is a common and simple way to do that, I > want to be notified by the document that it just did become the main > document. > This seems like a case where Apple (maybe even Mike) could look at extending NSDocumentController and NSDocument to support this. NSDocumentController obviously can tell when documents change... and I think it would be great if it could forward a Notification containing which document has become active, and which window controller if it has multiple windows for a document. I'm doing this manually now... it works... but a more 'standard' and supported method would be desirable.. From zmonster at heliosgroup.com Tue Mar 9 12:29:32 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Querying ODBC Setup for database server / database name Message-ID: <5948F7268092D211B2020000D11C0FB8081772@chi-serv2.heliosgroup.com> Is there a way in OpenStep 4.2 / NT (or Yellow Box on NT) to query the NT ODBC Control Panel in order to get the database server and database name for a given ODBC data source? If not, can stored procedure(s) be run to obtain the same result? If so, I assume these stored procedures would be database-specific... Eric From zmonster at heliosgroup.com Tue Mar 9 14:38:13 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:52 2005 Subject: Problems Printing Collated Pages Message-ID: <5948F7268092D211B2020000D11C0FB8081775@chi-serv2.heliosgroup.com> Has anyone ever experienced problems performing a print job with collated pages on OpenStep4.2 NT or Yellow Box for NT? A custom OpenStep app is not printing correctly (sometimes not at all) when the "collated" switch is clicked in the NT print panel. Just wondering if this might be a bug in OpenStep. Thank You, Eric From sanguish at digifix.com Tue Mar 9 21:08:15 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:52 2005 Subject: Localization on Windows NT Message-ID: <199903100506.AAA04606@digifix.com> On YB/NT how do you specify the language to use as far as localization goes? Can you? From frank at inis.de Wed Mar 10 04:10:19 1999 From: frank at inis.de (Frank Knobloch) Date: Thu Nov 3 14:41:52 2005 Subject: omnimake??? Message-ID: <199903101310.OAA02823@inis-9.inis.de> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 1185 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990310/5d04a42c/attachment.bin From paulrs at lgs-systems.com Wed Mar 10 06:38:52 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: No target/action methods for EODisplayGroup in IB Message-ID: <9903101438.AA10064@slab> RE: WOF 4.0 YB NT Hey Guys, Does anyone know why there are no target action methods defined in IB for EODisplayGroup. I have the EOF palette loaded, but when I try to connect a button to the 'insert:' method, the connection inspector shows that there are no methods available. The 'Classes' browser also reflects this. The only way I could get this to work (this was with a brand new nib), was to drag the .h for EODisplayGroup into the nib. What's up with this? Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From Malcolm_Crawford at plsys.co.uk Wed Mar 10 15:18:56 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:53 2005 Subject: No target/action methods for EODisplayGroup in IB In-Reply-To: <9903101438.AA10064@slab> References: <9903101438.AA10064@slab> Message-ID: <199903102318.XAA02481@jasmine.plsys.co.uk> Paul wrote: > Does anyone know why there are no target action methods defined in > IB for EODisplayGroup. I have the EOF palette loaded, but when I try to > connect a button to the 'insert:' method, the connection inspector shows > that there are no methods available. The 'Classes' browser also reflects > this. The only way I could get this to work (this was with a brand new > nib), was to drag the .h for EODisplayGroup into the nib. What's up with > this? > Umm, are you sure you have the right palette loaded? The one that shows the EOEditingContext and the DisplayGroup icons? The title of the left-hand column in the Connections Inspector should be a pop-up menu; it will allow you to choose which connections to show (e.g. EOAsssociations, Outlets etc.). One word of warning here: if you have an association set, don't start looking at a different set of connections, as the existing one will be broken (e.g. if you have an EOAssociation set, whilst inspecting it, don't switch to the Outlets column, else the EOAssociation is broken) -- this is easy not to spot, and you may wonder why your app stopped working... Best wishes, mmalc. From kgb at graduate.net Wed Mar 10 17:22:44 1999 From: kgb at graduate.net (KGB Software) Date: Thu Nov 3 14:41:53 2005 Subject: getting objects from TextFields with formatters? Message-ID: Hi, I have some NSTextFields with formatters on them. Some have NSDateFormatters, some have NSNumberFormatters. I'd like to get the fields' values as the appropriate corresponding objects (ie., NSCalendarDates or NSDecimalNumbers, respectively). How can I do this? Sending objectValue to the TextFields returns nil. I don't want to take the stringValue and convert it with NSCalendarDate's dateFromString:withFormat: method, since I'd have to change my source every time I change the format in IB. Any help? --John Kuszewski From zmonster at heliosgroup.com Wed Mar 10 19:59:32 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:53 2005 Subject: NSPrintPanel runModal Message-ID: <5948F7268092D211B2020000D11C0FB8081784@chi-serv2.heliosgroup.com> Hello, Has anyone had trouble with NSPrintPanel's 'runModal' method on NT? I wish to manually run a print panel and print operation, rather than call -[NSPrintOperation runOperation] (for various reasons). However, when I set up a new print operation, set the print panel on that operation to [NSPrintPanel printPanel], and then tell that print panel to 'runModal', I get a 'printer error'. An alert panel pops up (print panel never even makes it to the screen), and it says there is a 'printer error', and then the runModal methods returns with NSCancelButton status (and I didn't even get a chance to manipulate the print panel). What setup does NSPrintOperation's 'runOperation' method perform before running the print panel modal? It doesn't seem to be doing much, but it's doing something that I'm not doing which allows the print panel to display. Thanks, Eric From paulrs at lgs-systems.com Thu Mar 11 05:27:45 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: No target/action methods for EODisplayGroup in IB In-Reply-To: <199903102318.XAA02481@jasmine.plsys.co.uk> References: <199903102318.XAA02481@jasmine.plsys.co.uk> Message-ID: <9903111327.AA11070@slab> Malcolm wrote: [ Paul wrote: [ > Does anyone know why there are no target action methods defined in [ > IB for EODisplayGroup. I have the EOF palette loaded, but when I try [ > to connect a button to the 'insert:' method, the connection inspector [ > shows that there are no methods available. The 'Classes' browser also [ > reflects this. The only way I could get this to work (this was with a [ > brand new nib), was to drag the .h for EODisplayGroup into the nib. [ > What's up with this? [ > [ Umm, are you sure you have the right palette loaded? [ The one that shows the EOEditingContext and the DisplayGroup icons? Yes, I am sure. I've been working with EOF since 1.x days, so I'm not an idiot :-) Two others replied to me privately that they have experienced the same problem and the only remedy seemed to be quitting and restarting IB. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From Malcolm_Crawford at plsys.co.uk Thu Mar 11 07:11:59 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:53 2005 Subject: No target/action methods for EODisplayGroup in IB In-Reply-To: <9903111327.AA11070@slab> References: <199903102318.XAA02481@jasmine.plsys.co.uk> <9903111327.AA11070@slab> Message-ID: <199903111512.PAA02988@jasmine.plsys.co.uk> Paul wrote: > [ Umm, are you sure you have the right palette loaded? > [ The one that shows the EOEditingContext and the DisplayGroup icons? > > Yes, I am sure. I've been working with EOF since 1.x days, so I'm not an > idiot :-) > Hmm, I'm not sure there's a causal relationship there -- I've been playing with this stuff since before DBKit, but it doesn't stop me being an idiot. Or perhaps there was just a short window of preventative opportunity in 1995 or thenabouts! ;-) Best wishes, mmalc. From paulrs at lgs-systems.com Thu Mar 11 12:27:30 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: Sorting nibs in PB Message-ID: <9903112027.AA11501@slab> RE: YB NT, WOF 4.0 Hey guys, Does anyone know how to get ProjectBuilder to sort nibs from existing projects (ones created under OS 4.2) in alphabetical order? I have my preferences set to sort files by name, and this is working just fine for the class and header files, but I am having no luck with nibs. They just keep sorting based on file position. What gives? Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From paulrs at lgs-systems.com Thu Mar 11 12:41:52 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: Problem with NSCell's setFormatter: Message-ID: <9903112041.AA11566@slab> -------------- next part -------------- Skipped content of type multipart/mixed From sramp at tin.it Thu Mar 11 15:57:10 1999 From: sramp at tin.it (Stefano Rampazzo) Date: Thu Nov 3 14:41:53 2005 Subject: YellowBox for Windows - Only postscript printer ? Message-ID: Hello, please, I would like to know if there is any turnaround to fix the odd printer problem with the YellowBox for Windows. I mean about the fact that are supported only postscript printers. Thanks in advance Stefano From jkeenan at apple.com Thu Mar 11 16:48:13 1999 From: jkeenan at apple.com (Joe Keenan) Date: Thu Nov 3 14:41:53 2005 Subject: YellowBox for Windows - Only postscript printer ? References: Message-ID: <36E88E81.4E105788@apple.com> I use SuperPrint from Zenographics to set up a "fake" postscript printer. Works fine with OSE 4.2. joe Stefano Rampazzo wrote: > Hello, > > please, I would like to know if there is any turnaround to fix the odd > printer problem with the YellowBox for Windows. > I mean about the fact that are supported only postscript printers. > > Thanks in advance > Stefano From sanguish at digifix.com Thu Mar 11 23:31:45 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:53 2005 Subject: Revisiting HTMLEdit(or) - anyone interested? In-Reply-To: <9903112027.AA11501@slab> References: <9903112027.AA11501@slab> Message-ID: <199903120731.CAA09595@digifix.com> A few months back I stopped the HTMLEdit series on Stepwise for a variety of reasons including Apple's usurping of the HTMLEdit name for their own application. I'm not sure how to get back to the series when I basically have to go back and edit all the earlier files (and source) to reflect a name change. This hassle is DEFINATELY worth a couple of G3s donated to Stepwise from Apple BTW.. :-) The _internal_ copy of HTMLEditor has a number of features that the published article version doesn't have. - supports drag and drop of Netscape URLs on Windblows - converts basic RTF formatting to HTML This is a useful exercise in working with pasteboards, as well as reading RTF and walking through the structure of attributed strings. - converts pasted in graphics to GIF or PNG format shows how to write out GIF and PNG files using the NSImage and NSBitMapImageRep classes. There are still a couple of missing features: - conversion of actual text layout to HTML (adding

  • ) This adds alot of code and not related to YB specifically. If someone has some code that does this already available, I'd love to here about it. (I know that OmniFoundation does some of this.. but I don't want to have to rely on external code for it if possible.. adding all the frameworks to the simple app seems.. nasty) - HTML previewing in the app itself (with the new HTML support in YB/1.0 this is easily doable) So there are several options open, and this is where I need some input: 1. Release the version as is, no more articles 2. Write up articles for the currently finished features and release as is (which leaves us in a rather incomplete state) 3. Do 2, then add the HTML Previewing (with an article) and then add the conversion stuff without an article, release as source There are still more features to add (htmltidy is a cool little app that cleans your HTML code... it could easily be rolled into HTMLEditor) but I think anymore should be done outside of the confines of a series of articles (if you think otherwise.. let me know please!) There will be more articles on Stepwise of this nature.. I'm thinking about starting something new (documenting ColdCompress and releasing that source, or NicCentral and releasing that source).. From paulrs at lgs-systems.com Fri Mar 12 04:32:53 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: Crash with NSPopUpButton Message-ID: <9903121232.AA12166@slab> -------------- next part -------------- Skipped content of type multipart/mixed From wchin at acm.org Fri Mar 12 05:11:54 1999 From: wchin at acm.org (Bill Chin) Date: Thu Nov 3 14:41:53 2005 Subject: YellowBox for Windows - Only postscript printer ? Message-ID: <199903121311.IAA01145@nikita.eastwindtechnology.com> This really belongs in macosx-admin, so follow ups to this message should remove macosx-dev. A promising solution is using Ghostscript for Windows with RedMon, a print redirector for Windows. It's not perfect yet and I haven't had a chance to troubleshoot the problem. It prints text file output (ie. a MailViewer message w/o graphics) w/o a problem, but output with graphics causes Ghostscript to have fits. Interestingly enough, (and the reason why I think there is hope), if I print the postscript to file, open it with GSView, and then print it, Ghostscript doesn't have a problem. Printing postscript generated by Windows as opposed to YB works also. This is to a HP LaserJet 5L (using the 4L driver). See http://www.cs.wisc.edu./~ghost/ for Ghostscript and http://www.cs.wisc.edu./~ghost/redmon/index.html for RedMon. ..Bill Chin sramp@tin.it (Stefano Rampazzo) asked: > please, I would like to know if there is any turnaround to fix the odd > printer problem with the YellowBox for Windows. > I mean about the fact that are supported only postscript printers. From zmonster at heliosgroup.com Fri Mar 12 09:17:32 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:53 2005 Subject: NSPrintOperation / NSPrintPanel Bug Message-ID: <5948F7268092D211B2020000D11C0FB8081786@chi-serv2.heliosgroup.com> Hello, BUG PLATFORM: OpenStep 4.2 NT: I'm hoping someone could explain the following behavior, which seems to indicate that it is impossible to manually run a print panel: 1. This code does _not_ work (results in AppKit popping up an error panel): [[NSPrintPanel printPanel] runModal]; 2. This code does _not_ work (results in AppKit popping up an error panel): NSPrintOperation *op; op = [NSPrintOperation printOperationWithView:myView]; [op setPrintPanel:[NSPrintPanel printPanel]]; [[op printPanel] runModal]; 3. This code _does_ work: NSPrintOperation *op; op = [NSPrintOperation printOperationWithView:myView]; [op runOperation]; Therefore, 'runOperation' is doing something to initialize the NSPrintPanel in order to allow it to sucessfully display. I have set breakpoints on just about every NSPrintPanel/PrintOperation method, and I can't seem to identify the problem. Any ideas? Thank You, Eric From sanguish at digifix.com Fri Mar 12 15:18:53 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:53 2005 Subject: Hmmm... odd behaviour in NSBitMapImageRep Message-ID: <199903122316.SAA12065@digifix.com> This seems odd I'm loading a tiff image and displaying it in a row of an NSTableView. The odd thing is that it works with one tiff, but not with another.. and the tiff that doesn't work in this case, does work in every other application. Could this be some bizarre ColorSync interaction perhaps? (I'm guessing that because its fiddling with the dictionary, and as I recall, this dictionary is used for this sort of auxilliary behaviour. Is there a work around? (aside from messing with the tiff) Mar 12 18:13:23 xMediaEngine[6709] *** -[NSConcreteDictionary setObject:forKey:]: selector not recognized #6 0x43457b7c in -[NSBitmapImageRep(NSBitmapImageFileTypeExtensions) setProperty:withValue:] () #7 0x433392e8 in -[NSBitmapImageRep _loadImageInfoFromTIFF:] () #8 0x43353ea4 in -[NSBitmapImageRep _loadImageFromTIFF:imageNumber:] () #9 0x4335067c in -[NSBitmapImageRep _loadData] () #10 0x433a4498 in -[NSBitmapImageRep drawInRect:] () #11 0x43351ef4 in -[NSImage drawRepresentation:inRect:] () #12 0x4334f0a0 in -[NSImage _drawRepresentation:] () #13 0x43349050 in -[NSImage _cacheRepresentation:stayFocused:] () #14 0x433541b0 in -[NSImage _cacheRepresentation:] () #15 0x43335dc8 in -[NSImage _composite:delta:fromRect:toPoint:] () #16 0x433353cc in -[NSImage compositeToPoint:fromRect:operation:] () #17 0x433503a0 in -[NSImage compositeToPoint:operation:] () #18 0x433c59f0 in -[NSImageCell drawInteriorWithFrame:inView:] () #19 0x433ce770 in -[NSImageCell drawWithFrame:inView:] () #20 0x43424030 in -[NSTableView drawRow:clipRect:] () #21 0x43424cd4 in -[NSTableView drawRect:] () From jh at media.mit.edu Fri Mar 12 19:07:45 1999 From: jh at media.mit.edu (John Underkoffler) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: <199903130307.WAA00467@mediaone.net> Has anyone got the true & authentic contents of the NSApplication `run' method? Custom main event loops are much alluded to, but never divulged. We intend total domination, you see. Though of course a reliable hook into (out of, actually) the existing event loop at a known point in its cyclic execution would be workable too, likely. Bless you all, John From sanguish at digifix.com Fri Mar 12 21:44:35 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:53 2005 Subject: NSDocument - how to manage larget assets within a document wrapper? (rather URGENT) Message-ID: <199903130542.AAA13298@digifix.com> I've been having some issues with this for a while, and don't have a satisfactory solution that won't require substantial re-engineering.. so I'm hoping that someone else will have an idea. My app saves its self as a document wrapper. There are a cuople of small files, and then a number of large files, usually Images or QT movies. NSImages are easy because those classes seem to manage things quite nicely. Large images aren't loaded until you go to save. I can easily store a reference to an NSImage * in my object, and write it out easily as well. Quicktime movies however are not as easy. The problem here is that the NSMovieView ONLY accepts a path to the movie. So in this case, I need to manage the QT file contents myself. I could open a fileHandle on the QT file, and then I'm able to access the data, even if it moves (and you'll see why this is important). The problem is, that if the file moves, the path is no longer correct, and I can't load the movie. This really becomes an issue when you're trying to save the document. If there is a QT file in the application wrapper, you can't just track the filename because when you go to write out the file NSDocumentController has already moved the previously saved file to a new path name (file.wrapper~) and you're screwed. So in the broadest sense, what is the recommended method of handling large assets like this? I think NSFileWrapper actually loads all the data into memory at once, and regardless, you can't seem to get the path of the new file if the data moves.. I could open an NSFileHandle on the file, then I get access to the data, even if it moves, but I can't get the path name again (or am I missing something). From sanguish at digifix.com Sat Mar 13 00:13:14 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:53 2005 Subject: Is NSUserDefaults broken slightly? Message-ID: <199903130811.DAA13634@digifix.com> I'm trying to write a small command line tool to toggle the UnixExpert setting in NSGlobalDomain So, following the docs... myDefaults=[[NSUserDefaults alloc] init]; NSLog(@"setSearchList - %@",[myDefaults searchList]); [myDefaults setSearchList:[NSArray arrayWithObject:@"NSGlobalDomain"]]; NSLog(@"verify new setSearchList - %@",[myDefaults searchList]); unixExpertValue=[myDefaults boolForKey:@"NSUnixExpert"]; NSLog(@"Unix Expert: %@", unixExpertValue ? @"Yes" : @"No"); [myDefaults setBool:NO forKey:@"NSUnixExpert"]; NSLog(@"Unix Expert set to NO"); unixExpertValue=[myDefaults boolForKey:@"NSUnixExpert"]; NSLog(@"Unix Expert: %@", unixExpertValue ? @"Yes" : @"No"); success=[myDefaults synchronize]; if (success) NSLog(@"defaults were synchronized"); [myDefaults release]; However, for some odd reason, this is actually outputing this.. Mar 13 03:10:52 unixexpert[10758] setSearchList - () Mar 13 03:10:52 unixexpert[10758] verify new setSearchList - (NSGlobalDomain) Mar 13 03:10:52 unixexpert[10758] Unix Expert: Yes Mar 13 03:10:52 unixexpert[10758] Unix Expert set to NO Mar 13 03:10:52 unixexpert[10758] Unix Expert: Yes Mar 13 03:10:53 unixexpert[10758] defaults were synchronized And its writing the NSUnixExpert = NO to the domain for the tool (unixexpert) which isn't even part of the searchList. So... is this a bug? I've also tried using the standardUserDefaults class method, and this also works the same, which is contrary to the documentation (from the docs) A search for the value of a given default proceeds through the domains in an NSUserDefaults object's search list. Only domains in the search list are searched. The standard search list contains the domains from the table above, in the order listed. A search ends when the default is found. Thus, if multiple domains contain the same default, only the domain nearest the beginning of the search list provides the default's value. Using the setSearchList: method, you can reorder the default search list or set up one that is a subset of all the user's domains. Any comments? From bungi at omnigroup.com Sat Mar 13 00:27:07 1999 From: bungi at omnigroup.com (Timothy J. Wood) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: <199903130827.AAA03698@mungi.omnigroup.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 1108 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990313/f236ba2d/attachment.bin From sanguish at digifix.com Sat Mar 13 00:33:35 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:53 2005 Subject: NSUserDefaults - Workaround found... but the behaviour is still there Message-ID: <199903130831.DAA13724@digifix.com> Just so nobody wastes any time, I've worked around the specific issue I was trying to in the previous message (updating NSUnixExpert) by getting the PersistentDomain, removing it, changing the value in the dictionary, and then saving it again. But its not the way it should have to be done according to the docs... :-) From jh at media.mit.edu Sat Mar 13 09:45:41 1999 From: jh at media.mit.edu (John Underkoffler) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: <199903131745.MAA00338@mediaone.net> The point is that the event-loop model makes it extremely difficult to implement certain kinds of interaction structures; though a swell general-purpose approach, it's clearly inappropriate for times when a fella needs (for example) to do something or other exactly once per pass through -- and at a known point within -- the event loop. I'm sure you can see this. Some other not especially good environments start to get at this idea by way of an `idle' function; but that's not really the point. Threading's inappropriate here. NSTimers, similarly, don't cut it. So that requires a custom main event loop. Nothing really so hard about that. Again, therefore: has anyone got the accurate contents of one? Reiteratively, John From bungi at omnigroup.com Sat Mar 13 12:10:34 1999 From: bungi at omnigroup.com (Timothy J. Wood) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: <199903132010.MAA04416@mungi.omnigroup.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 2265 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990313/6b60f66b/attachment.bin From bbum at codefab.com Sat Mar 13 12:44:40 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop In-Reply-To: <199903131745.MAA00338@mediaone.net> Message-ID: The event loop model used by the AppKit is a hell of lot simpler than one might suspect. Because of the elegant object oriented design of the AppKit, it is possible to customize event handling in just about any way one could possibly want. NSApp's event loop simply: - sets up an autorelease pool - waits for an event - dispatches the event to the appropriate target - releases the the pool - repeat until dead This is exactly the same set of events that occur for any properly written yellow application that hangs out waiting for things to happen. When you launch the WOInfoCenter, it does something quite similar. As do service handlers and the workspace. All of these use NSRunLoop to congfigure the application for the pool of ports, timers, and other things that might produce events. Furthermore, all of the various event sources can be configured to only trigger the run loop when in certain modes. As an example, a yellow box application won't process workspace request events while a modal window is running. Check out this app: ftp://ftp.codefab.com/pub/unsupported/foo.tar.gz It is quite simple-- the NIB file for the main window contains random appkit objects, but they aren't connected to anything. The app contains a subclass of NSApplication as the app's application class. It also contains a subclass of NSRunLoop that poses as NSRunLoop upon load. The subclass of NSRunLoop prints out information about run loop manipulation performed by the application. You will see a surprising amount of activity-- most of this is caused by: - the application integrating itself into the rest of the system... i.e. setting itself up to be a target and client for services and the workspace, etc... - the application queries the postscript server to figure out the size of the screen, number of screens, when windows actually come into existance in the window server, etc... all of these occur asynchronously and the application configures the run loop to wait for the response on each. - the app sets itself up for various NSConnections that might be registered with it. When you actually bring the window to the front, you will see more activity on the run loop. - It installs an NSTimer that blinks the cursor on the first responder field - the event tracking entry starts tracking mouse motion and firing events in the run loop b.bum On Sat, 13 Mar 1999, John Underkoffler wrote: > The point is that the event-loop model makes > it extremely difficult to implement certain > kinds of interaction structures; though > a swell general-purpose approach, it's > clearly inappropriate for times when a fella > needs (for example) to do something or other > exactly once per pass through -- and at a known > point within -- the event loop. I'm sure you > can see this. > > Some other not especially good environments > start to get at this idea by way of an `idle' > function; but that's not really the point. > > Threading's inappropriate here. > > NSTimers, similarly, don't cut it. > > So that requires a custom main event loop. > Nothing really so hard about that. Again, > therefore: has anyone got the accurate > contents of one? > > Reiteratively, > John > > From yjongkind at fmco.com Sat Mar 13 12:45:33 1999 From: yjongkind at fmco.com (yonasj) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop References: <199903132010.MAA04416@mungi.omnigroup.com> Message-ID: <36EACE6D.1040E9B6@fmco.com> I am trying to dissect a similar problem. We are building a peice of software that processes many many messages that contain small peices of stock market data. We use threads to implement the underlying socket/msg model. These threads put messages into our application message queue. In an application with no GUI we do something like: while( 1 ) { msg = [MM hasMessage]; if( msg != nil ) [NSNotificationCenter postNotification....]; [NSRunLoop run]; } This causes problems when the application has a gui. We want to be able to send notifications to controllers, like "QuoteView" but with the model above, we have thread problems, because we can have only one thread displaying at a time. I was going to research how to solve this. Is it possible that maybe I could: -nextEventMatchingMask:untilDate:inMode:dequeue: { msg = [MM hasMessage]; if( msg != nil ) [NSNotificationCenter postNotification....]; [super nextEventMatchingMask:untilDate:inMode:dequeue:]; } -- Yonas. From marcel at system.de Sat Mar 13 12:01:33 1999 From: marcel at system.de (Marcel Weiher) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: > The point is that the event-loop model makes > it extremely difficult to implement certain > kinds of interaction structures; though > a swell general-purpose approach, it's > clearly inappropriate for times when a fella > needs (for example) to do something or other > exactly once per pass through -- and at a known > point within -- the event loop. I'm sure you > can see this. You can look at NSApplication's event handling methods which are called for each event if that is what you want to do. > So that requires a custom main event loop. > Nothing really so hard about that. Again, > therefore: has anyone got the accurate > contents of one? RTFM: look at the NSRunLoop documentation, specifically the -run method. Marcel From bbum at codefab.com Sat Mar 13 13:19:54 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop In-Reply-To: <36EACE6D.1040E9B6@fmco.com> Message-ID: If you want to move messages between threads, the traditional means has been to use mach port messages. Under Yellow box, you can use NSConnections and distributed objects (as they are OO wrappers around mach port messages). See the "Forming Connections Between Threads" section of NSConnection's documentation. To avoid the typical set of threading problems-- everything from memory corruption to dead locking to no performance advantage because of a bad threading model-- make sure and isolate objects into only one thread at a time. This proves to be quite easy with the NSConnection based model. When passing the object from the listen-for-quote-events thread to the main event loop thread, -retain the object before sending it from one thread to the other... release it on the other side. Actually, the best way to do this is to write a class whose sole purpose is to pass objects from one thread to another. When it receives the object ot be passed, it -retains it. Once the object successfully makes it to the other thread (and the other thread invokes the method to query for new objects), it returns the object -autoreleased. This effectively prevents the object from being -released while in between threads and maintains proper memory management symantics as the object hits the other thread. -- Of course, the question is.... is a multithreaded solution really appropriate here. I worked on a project that was quite similar; we had a quote server that served stock market data to many clients (70 to 80 clients at a time). The clients were single threaded open step apps. What we discovered was that multithreading in the client was not worth the trouble (multithreaded apps are notoriously difficult to design and develop. Maintenance/debugging can be pure hell. The end result often offers no real performance benefit.). Simply adding the port upon which stock update events arrived and treating update events like any other event provided more than enough performance to monitor 50+ individual stocks in a real-time trading environment. We also found that we could gain *a lot* of performance by focusing on effecient use of the network. Considering that every TCP/IP packet has 30 or 40 bytes of overhead when blasted across the 'net, we went to great lengths that the rest of the packet contained as much information as was possible-- we made sure that the server would effeciently queue updates and send updates for multiple symbols, if possible. Even with 70+ clients, the *server* wasn't even multithreaded! There was no need... the performance bottleneck was generally the network-- having multithreads trying to simultaneously write data to multiple clients didn't actually do *anythign* improve performance. The end result was that we had single threaded design throughout-- greatly reducing development and maintenance costs while greatly improving system reliability-- and still had a system that could handle updating 50+ different tickers across 70+ clients from a single server. b.bum On Sat, 13 Mar 1999, yonasj wrote: > I am trying to dissect a similar problem. > > We are building a peice of software that processes many many messages > that contain small peices of stock market data. We use threads to > implement the underlying socket/msg model. These threads put messages > into our application message queue. > > In an application with no GUI we do something like: > while( 1 ) > { > msg = [MM hasMessage]; > if( msg != nil ) > [NSNotificationCenter postNotification....]; > [NSRunLoop run]; > } > > This causes problems when the application has a gui. > > We want to be able to send notifications to controllers, like > "QuoteView" but with the model above, we have thread problems, because > we can have only one thread displaying at a time. > > I was going to research how to solve this. > > Is it possible that maybe I could: > > -nextEventMatchingMask:untilDate:inMode:dequeue: > { > msg = [MM hasMessage]; > if( msg != nil ) > [NSNotificationCenter postNotification....]; > [super nextEventMatchingMask:untilDate:inMode:dequeue:]; > > } > > > -- Yonas. > > From bungi at omnigroup.com Sat Mar 13 13:25:17 1999 From: bungi at omnigroup.com (Timothy J. Wood) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop Message-ID: <199903132125.NAA02133@mungi.omnigroup.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 2145 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990313/1a868fbd/attachment.bin From marcel at system.de Sat Mar 13 13:28:55 1999 From: marcel at system.de (Marcel Weiher) Date: Thu Nov 3 14:41:53 2005 Subject: Raising a timed exception? Message-ID: I am trying to raise an exception after a certain amount of time has passed (a timeout value). An obvious way of doing this is to set up an NSTimer, but this doesn't work because exceptions are ignored by NSTimers. Same goes for delayed message sends: exceptions are ignored. Both log the exception and the fact that they are ignoring it. I currently work around this by stashing the exception away in some private storage and manually checking for it later, but that is pretty ugly. Does anybody have an idea how I can keep the exception from being ignored? Marcel From dyoung at vviuh221.vvi.com Sat Mar 13 13:57:15 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop References: Message-ID: <9903132157.AA12423@vviuh221.vvi.com> > multithreading in the client was not worth the > trouble (multithreaded apps are notoriously difficult to design and > develop. Maintenance/debugging can be pure hell. The end result often > offers no real performance benefit.). That is definitely not our experiences. Multithreading is easy, efficient and well worth the effort for stock quote data as well as other sorts of data feeds. Some of our multithreaded apps collect data at 1000 events per seconds and display them in real time, others only a few times a second and with hundreds of tickers in a portfolio. I would highly recommend multi-threading for all sorts of data collection problems as well as queuing. Infact, it is built into some of our kits. > ... had single threaded design throughout-- greatly > reducing development and maintenance costs ... We have just the opposite experiences, multithreading simplified system design. Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From kelleysoft at pressenter.com Sat Mar 13 15:34:50 1999 From: kelleysoft at pressenter.com (Greg Betzel) Date: Thu Nov 3 14:41:53 2005 Subject: NSViews: removeFromSuperview & dealloc Message-ID: Hi, When you call removeFromSuperview on an NSView, there seems to be delay before the call takes effect & the view is actually released/deallocated (it appears the calling method has to exit first). Does anybody know if this is indeed correct, and if so, if there's a way to have the view removed & killed immediately? I imagine something like flushWindow: right after removing the view would work to sync everything, but I'd like to avoid that somehow. Thanks in advance. Greg Betzel From bbum at codefab.com Sat Mar 13 16:57:32 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:53 2005 Subject: seeking guts: main event loop In-Reply-To: <9903132157.AA12423@vviuh221.vvi.com> Message-ID: For data queuing applciations, multithreaded-- when applied carefully and appropriately-- can be quite useful. My message said as much and provided what I have found to be a very workable design. By carefully and appropriately, I mean that the threaded aspects of the application are isolated inside of a single class or mechanism that effectively hides the threaded aspects of the application. However, even with that in place, it is still very easy to screw up threading. Worse-- thread related bugs are typically extremely difficult to debug. Often, the difference between a debug and non-debug build or run can be enough to cause differences in thread scheduling that can lead to different symptoms when debugging thread related errors. Again-- threads have their place! As you have found, they can provide an elegant solution to a certain class of problems. But that solution comes with a certain set of requirements. Most importantly, it requires a thourough and complete design implemented by a group of developers with the discipline not to stray from that design. Equally as important, the design cannot change without an extremely careful analysis of how that affects the threaded aspects of the application. Even a properly built threaded solution has costs that have to be accounted for when weighing the value of multiple threads. The most significant cost is that of maintenance and upgrades. Unlike a non-threaded application, fixing defects in the behaviour of a multithreaded system must be done with extreme care-- if the defect is a plain old bug, it must be fixed within the bounds of the original design (in a single threaded system, there is considerable leeway in addressing bugs). If, however, the bug is the result of a defect in the original design, great care must be taken to ensure that fixing it does not incur very serious thread related problems. When upgrading an application-- adding new features, extending APIs, etc-- multi-threaded based systems often increase this cost greatly because the threading paradigm used is not ideal for the new requirements. --- The server in the stock information system that I mentioned in the previous message would often consume 1,000+ events per second. We found that by coalescing the incoming events into per client queues while round-robining out the updates to the clients, our single threaded server had no problem keeping up. As well, by placing this scheduling intelligence in the server we greatly reduced network traffic (which was important as some of the clients were on the other end of relatively slow links) and the clients were able to update the display more effeciently because they would receive a slew of update events simultaneously and could optimize the display accordingly. Clearly, we could make the server handle the load more effeciently with a multithreaded design. However, we found it cheaper simply to install faster server hardware-- the cost of maintaining a single threaded server was much less! Because we had other, less experienced with concurrent programming, developers using the frameworks, we concluded that a multithreaded solution would greatly increase the cost and risk of failure of their projects! When building codebases for use by other developers, a less than perfect implementation of threading with even slightly less than complete documentation can be deadly. Even if the code and docs are perfect and complete, multithreading can still greatly frustrate the third party developer-- if they slightly misinterpret either, it can lead to that wonderfully inconsistent and difficult to reproduce/debug set of bugs that can only be witnessed with a concurrent system. --- None of this is to say that I wouldn't use or recommend multiple thread solutions in the future. In my experience and research, I have found that threaded solutions can be much more costly to develop and maintain-- not all of the costs are obvious. I have also found that it is much more difficult to bring new developers up to speed on a project when it involves multiple threads. So few developers have worked with threaded systems-- much less threaded systems that are properly designed/built-- and, as such, most developers are somewhat uncomforatble with threaded codebases. Even if the developer is confident with their concurrent programming skills, if he or she has misinterpreted the specs and/or implementation, it is quite likely that the developer will introduce thread related bugs into the system that can be extremely difficult and frustrating to track down. I view threads as an optimization. Unless I can exactly quantify why threads will improve performance of an application before I start to build it, I won't use 'em. As with any software solution, it is often a costly waste of time to attempt to optimize the system before real world performance analysis has occurred. (I can't tell you how many times I have written or run across code where it was clear that I or someone else thought something was a performance bottleneck and spent loads of time making otherwise simple code comple through silly optimizations). This is especially true of well designed object oriented systems-- not only do OO systems make it possible to do concurrent programming with more grace and cleanliness than traditional functional systems-- they also allow one to build a bare bones implementation of the final solution without optimization to both prove that a concept works and to determine exactly where the bottlenecks are. In our modern computing environments-- ones where applications frequently must deal with events from many sources; time based, network based or user based events that occur in a completely unpredictable pattern-- multithreading is clearly more valuable than ever before. However, the increased complexity of the applications with the emphasis on a more open and programmable feature set means that there is more risk now than ever when deploying a multithreaded system. Success with threads cannot be maintained without careful design, a detailed implementation plan and a team of developers and managers that understand that they must follow the design and implementation plan to the letter. Of course, the same can be said of non threaded software solutions. Without a solid process and well documented plan throughout the project lifespan, it is inevitable that the project will eventually slip. Deadlines will be missed, deployments will be buggy, and the members of the project team will begin to burn out. b.bum On Sat, 13 Mar 1999, David Young wrote: > > > multithreading in the client was not worth the > > trouble (multithreaded apps are notoriously difficult to design and > > develop. Maintenance/debugging can be pure hell. The end result often > > offers no real performance benefit.). > > That is definitely not our experiences. Multithreading is easy, efficient > and well worth the effort for stock quote data as well as other sorts of data > feeds. Some of our multithreaded apps collect data at 1000 events per > seconds and display them in real time, others only a few times a second and > with hundreds of tickers in a portfolio. I would highly recommend > multi-threading for all sorts of data collection problems as well as queuing. > Infact, it is built into some of our kits. > > > ... had single threaded design throughout-- greatly > > reducing development and maintenance costs ... > > We have just the opposite experiences, multithreading simplified system design. > > Thanks A Bunch! David Young; VVI-DCS > http://www.vvi.com > > > > From marcel at system.de Sat Mar 13 16:08:21 1999 From: marcel at system.de (Marcel Weiher) Date: Thu Nov 3 14:41:53 2005 Subject: Raising a timed exception? In-Reply-To: Message-ID: I tried posting a notification without success. It had the same problem as the NSTimer / delayed execution mechanism because the notification gets executed from within the NSTimer / delayed execution method. Thanks, Marcel > From: Manu Iyengar > > You could post a notification instead, and have somebody who listens for > that notification raise the exception on your behalf? > > > >From the keyboard of Marcel Weiher on Sat, 13 Mar 1999: > > I am trying to raise an exception after a certain amount of time has > > passed (a timeout value). An obvious way of doing this is to set up > > an NSTimer, but this doesn't work because exceptions are ignored by > > NSTimers. Same goes for delayed message sends: exceptions are > > ignored. Both log the exception and the fact that they are ignoring > > it. > > > > I currently work around this by stashing the exception away in some > > private storage and manually checking for it later, but that is > > pretty ugly. Does anybody have an idea how I can keep the exception > > from being ignored? > > > > Marcel From paulrs at lgs-systems.com Sun Mar 14 04:45:40 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: Raising a timed exception? In-Reply-To: References: Message-ID: <9903141245.AA13366@slab> Marcel, How about doing this. Say that your timer fires some method like foo: - (void)methodToPerformAtTheTopOfTheNextEventLoop { // raise exception } - (void)foo { [self performSelector:@selector(methodToPerformAtTheTopOfTheNextEventLoop) withObject:nil afterDelay:0]; return; } Will that work? Regards, Paul Marcel wrote: [ I tried posting a notification without success. It had the same [ problem as the NSTimer / delayed execution mechanism because the [ notification gets executed from within the NSTimer / delayed [ execution method. [ [ Thanks, [ [ Marcel [ [ > From: Manu Iyengar [ > [ > You could post a notification instead, and have somebody who [ listens for [ > that notification raise the exception on your behalf? [ > [ > [ > >From the keyboard of Marcel Weiher on Sat, 13 Mar 1999: [ > > I am trying to raise an exception after a certain amount of time has [ > > passed (a timeout value). An obvious way of doing this is to set up [ > > an NSTimer, but this doesn't work because exceptions are ignored by [ > > NSTimers. Same goes for delayed message sends: exceptions are [ > > ignored. Both log the exception and the fact that they are ignoring [ > > it. [ > > [ > > I currently work around this by stashing the exception away in some [ > > private storage and manually checking for it later, but that is [ > > pretty ugly. Does anybody have an idea how I can keep the exception [ > > from being ignored? [ > > [ > > Marcel [ --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From alexnet at gestalt.com Sun Mar 14 19:44:00 1999 From: alexnet at gestalt.com (Alex Molochnikov) Date: Thu Nov 3 14:41:53 2005 Subject: Exposing ObjC classes to Java Message-ID: <9903150344.AA00612@gestalt.com> Did anyone master the art of exposing ObjC classes to Java? The YB documentation on JavaBridge is quite vague in that it does not explain how to package the generated Java class (actually, it barely explains anything at all). Here is what I have done: I created a simple ObjC class, Greeting, created a project of JavaWrapper type, and built it. The confusion sets right at this point: the documentation does not say what the expected outcome of building should be. I would expect something like a Java package that could be imported into the Java code that references the exposed ObjC class, but no... All that PB does is create a .dll and .lib file that presumably should be used when building the hybrid ObjC-Java app. To be sure, the documentation says that PB will generate a .java file for the ObjC class, a C stubs file, and a file with the initialization code for ObjC classes, but it does not say how they should be used, if at all. Worse yet, these files are buried in the .build directory tree suggesting that they are not for the user's consumption. Now, the Java code that references the wrapped ObjC class cannot find its definition because there is no package to import that would carry the wrapped class. I tried to force it by bringing the generated Greeting.java file from the JavaWrapper project into the test case, and this made the Java compiler happy... but not the Java VM. It now complains that it cannot find Greetings class, even though I brought the generated .dll library into the app environment. Any clues will be greatly appreciated. Thanks in advance. Alex Molochnikov alex@gestalt.com From bruce at iterative.com Sun Mar 14 21:35:20 1999 From: bruce at iterative.com (Bruce Fancher) Date: Thu Nov 3 14:41:53 2005 Subject: Exposing ObjC classes to Java In-Reply-To: <9903150344.AA00612@gestalt.com> Message-ID: <000101be6ea5$9dd95dc0$02e9e90a@deadlord> This doesn't exactly solve your problem, but what I have a Java class which I need to create and insert EO's. However, instead of wrapping the EO classes, what I've done is to just use valueForKey() and takeValueForKey() to call methods on the EO's, including methods other than accessors. So basically if I want to call the foo method on bar I just do bar.valueForKey("foo"). If I have a method which needs to accept a parameter I just call it e.g. - setFoo:, then call bar.takeValueForKey(value, "foo"). Note that this is approach is a terrible kludge, and I'm only doing it because this is a prototype which I'm going to use to seek funding and getting it up and running is more important than elegance at this stage in the process. But, it does the job and may be useful for you as well. Bruce > -----Original Message----- > From: macosx-dev@omnigroup.com [mailto:macosx-dev@omnigroup.com]On > Behalf Of Alex Molochnikov > Sent: Sunday, March 14, 1999 10:55 PM > To: Multiple recipients of list > Subject: Exposing ObjC classes to Java > > > Did anyone master the art of exposing ObjC classes to Java? The > YB documentation on JavaBridge is quite vague in that it does not > explain how to package the generated Java class (actually, it > barely explains anything at all). > > Here is what I have done: > > I created a simple ObjC class, Greeting, created a project of > JavaWrapper type, and built it. The confusion sets right at this > point: the documentation does not say what the expected outcome > of building should be. I would expect something like a Java > package that could be imported into the Java code that references > the exposed ObjC class, but no... All that PB does is create a > .dll and .lib file that presumably should be used when building > the hybrid ObjC-Java app. To be sure, the documentation says that > PB will generate a .java file for the ObjC class, a C stubs file, > and a file with the initialization code for ObjC classes, but it > does not say how they should be used, if at all. Worse yet, these > files are buried in the .build directory tree suggesting that > they are not for the user's consumption. > > Now, the Java code that references the wrapped ObjC class cannot > find its definition because there is no package to import that > would carry the wrapped class. > > I tried to force it by bringing the generated Greeting.java file > from the JavaWrapper project into the test case, and this made > the Java compiler happy... but not the Java VM. It now complains > that it cannot find Greetings class, even though I brought the > generated .dll library into the app environment. > > Any clues will be greatly appreciated. > > Thanks in advance. > > Alex Molochnikov > alex@gestalt.com From paulrs at lgs-systems.com Mon Mar 15 05:53:04 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: IB Intances Outline View Message-ID: <9903151353.AA13852@slab> Hey guys, I posted on this last week and got zero answers. I think the post didn't get through to the list. For some reason, the new IB (the one included with WOF 4.0) does not show the name that you have assigned to an instance of an object in the outline view. I find it very difficult to work in icon view, but it's now virtually impossible to work in outline view. Does anyone know if this is a bug or an intended change? If it is an intended change, does anyone know of a workaround? If not, does anyone know of a version of class-dump which works on NT so that I can muck around on my own and try to find a hook to put the information back in? Finally, can anyone verify that this is the state of affairs in the latest build of Mac OS X Server? Any suggestions would be appreciated. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From maury at OAAI.COM Mon Mar 15 07:26:39 1999 From: maury at OAAI.COM (Maury Markowitz) Date: Thu Nov 3 14:41:53 2005 Subject: Advice on setting file permissions Message-ID: <199903151516.KAA29570@OAAI.COM> While porting back to OS4.2 it was noticed that documents from our app are being written out with full permissions turned on. I check on my (DR2) side and the same thing is true. Now changing this seems easy enough from a coding perspective, but I'm not convinced I know exactly what I should be changing it to. Is there any sort of standard permission behaviour we should be following? If not, does anyone have any rule-of-thumb type guidelines? Maury From sam at darth.geg.mot.com Mon Mar 15 12:32:36 1999 From: sam at darth.geg.mot.com (Sam M. Daniel) Date: Thu Nov 3 14:41:53 2005 Subject: Tkl-Tk on MacOSX? Message-ID: <9903152032.AA03385@darth.geg.mot.com> Some of you may be quite interested to know that the multi-paradigm language Mozart (http://www.mozart-oz.org) has been essentially ported to RDR2 and will hopefully be ported to MacOSX Server very soon. However, without Tkl-Tk and Emacs, important functionality will be missing. Are Tkl-Tk and Emacs available on RDR2 or MacOSX? Were they not available on OS 4.2? Is X-Windows on MacOSX Server the only option currently and, if so, where does obtain a copy? Sam Daniel Motorola Inc. 602-441-7431 From alexnet at gestalt.com Mon Mar 15 11:42:18 1999 From: alexnet at gestalt.com (Alex Molochnikov) Date: Thu Nov 3 14:41:53 2005 Subject: Exposing ObjC classes to Java Message-ID: <9903151942.AA00411@gestalt.com> My sincere thanks to everyone who responded to my original post. I have another question related to the first one. While struggling with the problem of exposing ObjC to Java, I stumbled upon /Apple/Developer/Java/Headers/java-objc.h file. In its description of NSCreateJavaWrapper() function that as far as I can tell is at the center of the ObjC-Java bridge, if found the following ominous warning: * Note also that this function is extremely very not thread safe! If one develops a multi-threaded server that is made of both ObjC and Java, and ObjC classes are called from different Java threads, then the bridge will collapse, won't it? OTOH the Yellow Box frameworks are officially declared thread-safe, and they are exposed to Java through the same bridge technology. Go figure... Can anyone shed some light on this puzzle? Thanks again. Alex Molochnikov alex@gestalt.com From kc at omnigroup.com Mon Mar 15 12:09:14 1999 From: kc at omnigroup.com (Ken Case) Date: Thu Nov 3 14:41:53 2005 Subject: Advice on setting file permissions Message-ID: <199903152009.MAA06849@ignem.omnigroup.com> > While porting back to OS4.2 it was noticed that documents from > our app are being written out with full permissions turned on. > I check on my (DR2) side and the same thing is true. Now changing > this seems easy enough from a coding perspective, but I'm not > convinced I know exactly what I should be changing it to. What you want to do is create a file which obeys the "File-Creation Mask" settings in Expert Preferences. To do this, you should grab the file creation mode mask (see "man umask"), and mask your permissions against it. The easy way to do all of this is to grab the relevant source code from OmniFoundation: the stuff you want is in the category NSFileManager (OFExtensions): - (NSNumber *)defaultFilePermissions; - (NSNumber *)defaultDirectoryPermissions; Also note that OPENSTEP/Mach 4.2 had a bug where -[NSFileManager createFileAtPath:contents:attributes:] with nil attributes would create a file with the permissions of 0700 (u=rwx), rather than 0666 (a=rw) modified by the umask (which should be the default set of permissions). Thus, all files created using that call (by OmniWeb 3.0 beta 8, for example) were made executable, which meant that when opening them in the Workspace tried to execute them rather than view them. There's a OPENSTEP-only patch to fix this behavior in the category NSFileManager (OFBugFix) in OmniFoundation. Which reminds me: in Rhapsody DR2 (and perhaps Mac OS X Server) there's a bug where -[NSFileWrapper writeToFile:atomically:updateFilenames:] ignores the umask when creating files. (You can test this by running TextEdit, switching to Rich Text mode, dragging in a file or image so it creates an RTFD file, and saving to /tmp/foo.rtfd or whatever. You'll find that this newly created document is writable by everyone on the system.) In OmniAppKit there's a patch to fix this behavior in the category NSFileWrapper (OAExtensions). The source code to OmniFoundation and OmniAppKit are freely available from our developer resources page at http://www.omnigroup.com/DeveloperResources/. Hope this helps! Ken From paulrs at lgs-systems.com Mon Mar 15 17:13:31 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:53 2005 Subject: NSSplitView problem? Message-ID: <9903160113.AA14436@slab> James Klett wrote: [ Has anyone else noticed this wierd quirk of the NSSplitView?:: [ [ In IB when selecting two or more Views/Box's an grouping them [ in a SplitView, sometimes IB re-orders the views so that the one [ on the bottom becomes the one on the top. And sometimes when [ it orders them correctly, it will change the order right in front of [ your eyes when you click into one of the boxes to edit fields in the [ interior. [ [ I have been unable to figure out the rule for why IB does this sometimes [ and not other times. If anyone knows please let me know [ Jim et al, We're having a similar problem. In fact, in one situation with which I am currently dealing, I cannot get the split view to maintain the order of my views no matter what I do! This is extremely frustrating. If anyone knows of a workaround, please post it! I have had to resort to creating the split view on OS 4.2 and then bringing the nib over. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From cwhite at comnetix.com Mon Mar 15 23:25:10 1999 From: cwhite at comnetix.com (Craig White) Date: Thu Nov 3 14:41:53 2005 Subject: Inter thread communication with NSConnection Message-ID: <9903160225.AA106272@deli> I'm stumped on this one... any help is appreciated. In OpenStep 4.2 on Windows, I'm using an NSConnection between two threads (similar to the example in the docs), and despite the use of a protocol indicating my method is (oneway void), my calling thread will block if the other thread is busy! I can't seem to get the NSConnection to act as a queue for messages between the threads. What I want to achieve is asynchronous communication between threads (no blocking!), and control over when the receiving thread processes incoming messages (ie. with run-loop modes). Has anyone accomplished this with or without NSConnection? Craig From eric at alum.mit.edu Tue Mar 16 10:00:51 1999 From: eric at alum.mit.edu (Eric Hermanson) Date: Thu Nov 3 14:41:53 2005 Subject: Name Mangling in Objective-C++ Message-ID: <014b01be6fd6$edf8e9b0$5701020a@chi-prog-10.heliosgroup.com> We're having trouble linking an Objective-C++ framework that calls API included in a C++ third-party library (that was assumedly created with Microsoft's development tools). We get a bunch of undefined/redefined symbols link errors. It appears at first glance that the Objective-C++ compiler can't understand the "mangled" C++ function and class names. Are there compiler/linker flags that must be set when linking a C++ dynamic library (other than the -Obj-C++ flag)? I can't seem to find any information about this topic in the archives. Thank You, Eric From allan at alitech.com Tue Mar 16 02:01:25 1999 From: allan at alitech.com (Allan Noordvyk) Date: Thu Nov 3 14:41:53 2005 Subject: Name Mangling in Objective-C++ Message-ID: <00c201be6f93$f498e8d0$3001010a@nemo.ali.bc.ca> Eric Hermanson wrote: >We're having trouble linking an Objective-C++ framework that calls API >included in a C++ third-party library (that was assumedly created with >Microsoft's development tools). We get a bunch of undefined/redefined >symbols link errors. It appears at first glance that the Objective-C++ >compiler can't understand the "mangled" C++ function and class names. > >Are there compiler/linker flags that must be set when linking a C++ dynamic >library (other than the -Obj-C++ flag)? I can't seem to find any >information about this topic in the archives. There is no C++ name mangling standard. As such you can not usually call C++ objects compiled by one compiler (e.g. MS's VC++) from code compiled with a different compiler (e.g. NeXT/Apple's gcc). The only real work-around is to define a C function API which is wrapped around an OO stuff on either side, since C symbol construction is standardized between vendors. We have used this extensively to link MFC and OpenStep-based applications and libraries and it works reliably. -- Allan Noordvyk, Software Artisan ALI Technologies From cmh at greendragon.com Tue Mar 16 14:13:14 1999 From: cmh at greendragon.com (Chris Hanson) Date: Thu Nov 3 14:41:53 2005 Subject: Name Mangling in Objective-C++ In-Reply-To: <00c201be6f93$f498e8d0$3001010a@nemo.ali.bc.ca> Message-ID: At 12:33 PM -0600 3/16/99, Allan Noordvyk wrote: >There is no C++ name mangling standard. As such you can not usually >call C++ objects compiled by one compiler (e.g. MS's VC++) from code >compiled with a different compiler (e.g. NeXT/Apple's gcc). I thought there was a C++ binary standard on Windows, defined by Microsoft and Visual C++. Does gcc not conform to it? If it doesn't, when is conformance planned? From mtarbell at tarbell.jpl.nasa.gov Tue Mar 16 15:57:29 1999 From: mtarbell at tarbell.jpl.nasa.gov (Mark Tarbell) Date: Thu Nov 3 14:41:53 2005 Subject: Name Mangling in Objective-C++ In-Reply-To: References: Message-ID: <199903162357.AA01062@tarbell.jpl.nasa.gov> Chris Hanson wrote: > I thought there was a C++ binary standard on Windows, defined by > Microsoft and Visual C++. Does gcc not conform to it? If it > doesn't, when is conformance planned? I went through this using GNU and Sun's CC compiler. According to what I was told by Sun reps, their C++ name mangling algorithm is proprietary, due to it not being specified in the C++ specification. Apparently this is why GNU doesn't have an option to generate Windows or any other compatible name mangling: the mangling isn't defined in any standard. Microsoft doesn't define the C++ standard. Templates are another area where implementation is largely unspecified by the standard (meaning it's up to the individual manufacturers). There's no reason why there couldn't be a standard for mangling, though, except that it's probably too late to introduce one now. -- Mark Tarbell (Mark.A.Tarbell@jpl.nasa.gov) Applications Development Section Information Systems Development and Operations Division Engineering and Science Directorate Jet Propulsion Laboratory 4800 Oak Grove Drive MS 502-500 Pasadena, CA 91109 Opinions expressed herein do not reflect those of JPL. From jhendry at cmg.FCNBD.COM Tue Mar 16 17:54:06 1999 From: jhendry at cmg.FCNBD.COM (Jonathan Hendry) Date: Thu Nov 3 14:41:53 2005 Subject: YellowBox/NT? Message-ID: <199903170154.TAA23077@wo1085.cmg.FCNBD.COM> So- any word on YellowBox NT? Obviously, it's part of WebObjects. But can anyone *else* ship YB apps? From erik at object-factory.com Wed Mar 17 01:58:56 1999 From: erik at object-factory.com (Erik Doernenburg) Date: Thu Nov 3 14:41:53 2005 Subject: Inter thread communication with NSConnection Message-ID: <199903170958.BAA20283@ignem.omnigroup.com> > I'm using an NSConnection between two threads (similar to the example in > the docs), and despite the use of a protocol indicating my method is (oneway > void), my calling thread will block if the other thread is busy! Did you use 'setProtocolForProxy:' on the proxy of the object that implements the oneway void message? If you don't the NSConnection will do a synchronous check to see whether the target implements the selector before invoking it. regards, erik -- OBJECT FACTORY, Gesellschaft fuer Informatik und Datenverarbeitung mbH Telephone: ++49 +231 9751370 . Internet: http://www.object-factory.com From dyoung at vviuh221.vvi.com Wed Mar 17 02:52:31 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:53 2005 Subject: YellowBox/NT? References: <199903170154.TAA23077@wo1085.cmg.FCNBD.COM> Message-ID: <9903171052.AA04873@vviuh221.vvi.com> jhendry@cmg.FCNBD.COM asked: > So- any word on YellowBox NT? > Obviously, it's part of WebObjects. But can anyone *else* ship YB apps? On Wed, 2 Dec 1998 06:43:49 -0800 (PST) Alex Horovitz said: > I am pleased to be able to tell you that for all intents and > purposes the legal issues around licensing of the YB runtime for > windows are behind us. What remains before we will make a public > announcement of the licensing structures is agreement by the pricing > committee to the new pricing structure. I do expect to be able to > give you all a timeline in the near future, and I expect that Apple > will issue a press release at the end of that timeline period with > the details of the licensing arrangement. > ... That being said, I ask for your continued patience as approval of > pricing can sometimes take a little while ... Alex, what is the "near future" and "a little while"? Is it before the year 2000? On Wed, 25 Nov 1998 17:58:10 -0800 (PST) Alex Horovitz said: > ... Your support of our products and commitment to > excellence are something that we at Apple are all very thankful for. Thankful enough to actually come through on your promises? Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From xavierb at proxyware.fr Wed Mar 17 02:48:12 1999 From: xavierb at proxyware.fr (Xavier Barrier) Date: Thu Nov 3 14:41:53 2005 Subject: Inter thread communication with NSConnection Message-ID: <9903171048.AA03981@proxyware.fr> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 1544 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990317/fbfbf9e3/attachment.bin From nberch at db.lv Wed Mar 17 03:14:08 1999 From: nberch at db.lv (Nils Berzins) Date: Thu Nov 3 14:41:53 2005 Subject: Changes in Yellow box... Message-ID: <199903171116.NAA18289@saulite> Hi, Now, when the Mac OS X Server is finally released (at least Macweek and Stepwise claims so), it would be very nice if someone could put together a FAQ (or something like that) about what's new and what's changed in Yellow APIs since DR2. I suspect that it will take weeks before anyone unfortunate enough to live outside the US will get their hands on Mac OSXS... Happy St. Patrick's day, Nils Berzins P.S. It's strange, but I still don't see any announcemts in www.apple.com as well as no Yellow topics in wwdc99. :-( From xavierb at proxyware.fr Wed Mar 17 03:35:44 1999 From: xavierb at proxyware.fr (Xavier Barrier) Date: Thu Nov 3 14:41:53 2005 Subject: Changes in Yellow box... In-Reply-To: <199903171116.NAA18289@saulite> References: <199903171116.NAA18289@saulite> Message-ID: <9903171135.AA04740@proxyware.fr> You wrote: > Hi, > > Now, when the Mac OS X Server is finally released (at least Macweek and > Stepwise claims so), it would be very nice if someone could put together > a FAQ (or something like that) about what's new and what's changed in > Yellow APIs since DR2. I suspect that it will take weeks before anyone > unfortunate enough to live outside the US will get their hands on Mac > OSXS... > > Happy St. Patrick's day, > Nils Berzins > > P.S. It's strange, but I still don't see any announcemts in www.apple.com > as well as no Yellow topics in wwdc99. [An attachment was originally included here] > > Please take a closer look to http://www.apple.com. You can't miss it, it's right in the middle including retail price! :-) Regarding YB/NT, i did not find any information regarding any priceList. --- Xavier BARRIER --------------------------------------------------------- ProxyWare SARL 8, rue de l'Isly 75008 Paris France Financial Software Company Tel : (+33) 1 44 69 83 06 Fax : (+33) 1 44 69 06 92 e-mail : xavierb@proxyware.fr -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 1033 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990317/7e76e6c2/attachment.bin From sbrandon at music.gla.ac.uk Wed Mar 17 03:46:34 1999 From: sbrandon at music.gla.ac.uk (Stephen Brandon - SysAdmin) Date: Thu Nov 3 14:41:53 2005 Subject: SoundKit Open Source!!!!!! Message-ID: <199903171146.LAA13098@music.gla.ac.uk> Hi, Just as I was about to finish and release my SndKit, a free open sourced replacement for SoundKit, SoundKit goes OpenSource. Typical. I've spent a lot of time on it. Still, WOW! Fantastic news! See those SoundKit bugs disappear before your very eyes! I am so sad that I'm in the middle of purchasing WinNT machines with WebObjects (for the dev tools) as a replacement for our NeXT machines... :-( But MacOSX Server has come too late (and also would require ATM networking support, for our network). Oh the joys. I am both deliriously happy about SoundKit and NetInfo, and really upset at the fact we can't make use of the new OS. Stephen Brandon --- Systems Administrator, Department of Music, e-mail: S.Brandon@music.gla.ac.uk 14 University Gardens, (NeXT mail welcomed) University of Glasgow, Tel: +44 (0)141 330 6065 Glasgow. Fax: +44 (0)141 330 3518 Scotland G12 8QH From boerny at xenia.hsh.stusta.mhn.de Wed Mar 17 05:19:07 1999 From: boerny at xenia.hsh.stusta.mhn.de (Bernhard Scholz) Date: Thu Nov 3 14:41:53 2005 Subject: Compatibility of Rhapsody > MacOSX-Server In-Reply-To: References: Message-ID: <199903171319.OAA01493@xenia.hsh.stusta.mhn.de> Hello, Because we lack a G3 with MacOSX Server we currently can't figure out what to do with our Rhapsody software on the Peanuts Archive ( http://www.peanuts.org/ ). We are currently thinking of including the Rhapsody stuff in a new MacOSX Server directory or just to rename it. The later would imply that compiled Rhapsody software is able to run out-of-the-box on MacOS-X Server. Is this the case? Are Rhapsody binaries compatible with MacOSX Server (as long as there is a PPC version) and does Rhapsody Source Code compile out-of-the-box on MacOS-X Server? Any advice on what to do would be greatly appreciated. Greetings, Bernhard. --- Bernhard Scholz http://www.peanuts.org/ Peanuts-Archive Admin scholz@peanuts.org From bbum at codefab.com Wed Mar 17 06:02:00 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:53 2005 Subject: Compatibility of Rhapsody > MacOSX-Server In-Reply-To: <199903171319.OAA01493@xenia.hsh.stusta.mhn.de> Message-ID: Not sure what you are asking here.... .... Mac OS X Server is Rhapsody. They are the same thing. Rhapsody was the nickname for Mac OS X Server development. The binaries are mostly compatible. By mostly, I mean that some apps will exhibit instability or other weirdness because RDR2 was a pre-release version of Mac OS X Server and, as such, things have changed slightly in between the two versions. b.bum On Wed, 17 Mar 1999, Bernhard Scholz wrote: > Hello, > > Because we lack a G3 with MacOSX Server we currently can't figure > out what to do with our Rhapsody software on the Peanuts Archive > ( http://www.peanuts.org/ ). > > We are currently thinking of including the Rhapsody stuff in a new > MacOSX Server directory or just to rename it. The later would imply > that compiled Rhapsody software is able to run out-of-the-box on > MacOS-X Server. Is this the case? > > Are Rhapsody binaries compatible with MacOSX Server (as long as there is > a PPC version) and does Rhapsody Source Code compile out-of-the-box > on MacOS-X Server? > > Any advice on what to do would be greatly appreciated. > > Greetings, > > Bernhard. > > --- > Bernhard Scholz http://www.peanuts.org/ > Peanuts-Archive Admin scholz@peanuts.org > > From jes at rednsi.com Wed Mar 17 07:40:13 1999 From: jes at rednsi.com (Josep Egea i Sanchez) Date: Thu Nov 3 14:41:53 2005 Subject: Does MacOS X Server include an Oracle Adaptor Message-ID: <9903171540.AA07972@rednsi.com> Hi, The subject says it all. My company (and probably many others) would really appreciate if any of the lucky guys who get to see the box in the early future can comment on this matter. Thanks and best regards --- Josep Egea - jes@rednsi.com - NeXTMail & MIME OK NEXUS Servicios de Informacion - Barcelona (Spain) From marco at sente.ch Wed Mar 17 08:06:54 1999 From: marco at sente.ch (Marco Scheurer) Date: Thu Nov 3 14:41:53 2005 Subject: cvs and project builder In-Reply-To: <199902041517.QAA00442@xlthlx.sente.ch> References: <199902041517.QAA00442@xlthlx.sente.ch> Message-ID: <199903171607.RAA12264@hnw.sente.ch> On February 4, I wrote: > On Jan 27, 1999, Steve Loranz wrote: > > > [...] It would be great to see CVL provide services for other > > apps to utilize > > It does! In fact it provides a single service, to select, update > and display the status of a given file. This should cover 90% of > all needs. After activating the service you're in CVL, so you can > commit, update, etc. We decided against providing other cvs > operations as services, because you usually wants to know the > status of a file before doing anything with it. > > Unfortunately, last time I checked, ProjectBuilder does not put > correct file path information on the service pasteboard, so this > CVL service is unusable from PB ... A small ProjectBuilder bundle that fixes that behavior is available as source code from http://www.sente.ch/software/cvl/ . I do not know yet if the latest release of ProjectBuilder, from MacOSX Server, needs this patch. Marco Scheurer Sen:te From zander at rkinc.com Wed Mar 17 08:11:53 1999 From: zander at rkinc.com (Aleksey Sudakov) Date: Thu Nov 3 14:41:53 2005 Subject: Swing Message-ID: <9903171611.AA01068@rkinc.com> Hello, MacOSX Server comes with the JDK 1.1.6. It's quite outdated. DR2 also came with JDK 1.1.6 which was dog slow and as a result could not be used to run any Swing applications. Thus the question to anyone who have got MacOSX server already: How good is Java performance and is there anybody who successfully run any Swing/JFC applications? Thanks, Aleksey From pierce at twinforces.com Wed Mar 17 08:15:33 1999 From: pierce at twinforces.com (Pierce T. Wetter III) Date: Thu Nov 3 14:41:53 2005 Subject: Does MacOS X Server include an Oracle Adaptor In-Reply-To: <9903171540.AA07972@rednsi.com> Message-ID: At 7:48 AM -0800 3/17/99, Josep Egea i Sanchez wrote: > Hi, > > The subject says it all. My company (and probably many others) would really > appreciate if any of the lucky guys who get to see the box in the early > future can comment on this matter. Don't have the box, but one of Apples descriptions says that Oracle and ODBC is "coming". Someone has to port the Oracle libraries to MOSXS, for the adaptor to work, and ODBC probably requires an ODBC for UNIX library like that sold by various people. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From afindlay at austin.rr.com Wed Mar 17 08:22:29 1999 From: afindlay at austin.rr.com (adhamh findlay) Date: Thu Nov 3 14:41:53 2005 Subject: Does MacOS X Server include an Oracle Adaptor In-Reply-To: <9903171540.AA07972@rednsi.com> Message-ID: At 7:48 AM -0800 3/17/99, Josep Egea i Sanchez wrote: >Hi, > >The subject says it all. My company (and probably many others) would really >appreciate if any of the lucky guys who get to see the box in the early >future can comment on this matter. > >Thanks and best regards >--- >Josep Egea - jes@rednsi.com - NeXTMail & MIME OK >NEXUS Servicios de Informacion - Barcelona (Spain) It does not ship with one. adhamh From jes at rednsi.com Wed Mar 17 08:24:00 1999 From: jes at rednsi.com (Josep Egea i Sanchez) Date: Thu Nov 3 14:41:53 2005 Subject: [SUMMARY] Does MacOS X Server include an Oracle Adaptor Message-ID: <9903171624.AA08184@rednsi.com> Hi again, According to several responses I received, MacOS X Server ships without Oracle and Sybase adaptors, but both are expected to appear in the near future. I can only add that an Oracle adaptor is a "must have" for my company to consider buying the OS (along with Apple hardware). So we'll be waiting... Thanks and best regards --- Josep Egea - jes@rednsi.com - NeXTMail & MIME OK NEXUS Servicios de Informacion - Barcelona (Spain) Begin forwarded message: Date: Wed, 17 Mar 1999 07:45:55 -0800 (PST) From: Josep Egea i Sanchez To: Multiple recipients of list Subject: Does MacOS X Server include an Oracle Adaptor Hi, The subject says it all. My company (and probably many others) would really appreciate if any of the lucky guys who get to see the box in the early future can comment on this matter. Thanks and best regards --- Josep Egea - jes@rednsi.com - NeXTMail & MIME OK NEXUS Servicios de Informacion - Barcelona (Spain) From markm at mail.tyrell.com Wed Mar 17 08:27:01 1999 From: markm at mail.tyrell.com (Mark F. Murphy) Date: Thu Nov 3 14:41:53 2005 Subject: Mac OS X Server FAQ - MOSXS <> Rhapsody In-Reply-To: <9903171135.AA04740@proxyware.fr> Message-ID: Gotta love this: Q. Is Mac OS X Server the same as Rhapsody? A. No. Mac OS X Server leverages a number of the technologies formerly called Rhapsody. In addition, it includes some innovative services that were not part of the original Rhapsody project. mark --------------------------------------------------------------------------- Mark F. Murphy, Director Software Development Tyrell Software Corp PowerPerl(tm), Add Power To Your Webpage! --------------------------------------------------------------------------- Member of Families Against Internet Censorship: rainbow.rmi.net/~fagin/faic From hunter at lastonepicked.com Wed Mar 17 08:29:57 1999 From: hunter at lastonepicked.com (Hunter Hillegas) Date: Thu Nov 3 14:41:53 2005 Subject: Swing Message-ID: <19990317163120383.AAA474@nts1.webhostingprovider.com@firewall.jacobstern.com> Yeah, that's a great question... 1.1.6 isn't THAT outdated. The last iteration for 1.1 was 1.1.7 (right?) and 1.2 isn't really accepted very widely yet... Has anyone gotten their shrinkwrapped MacOS X Server package? ---------- >From: Aleksey Sudakov >To: Multiple recipients of list >Subject: Swing >Date: Wed, Mar 17, 1999, 8:23 AM > > Hello, > > MacOSX Server comes with the JDK 1.1.6. It's quite outdated. DR2 also came > with JDK 1.1.6 which was dog slow and as a result could not be used to run > any Swing applications. Thus the question to anyone who have got MacOSX > server already: How good is Java performance and is there anybody who > successfully run any Swing/JFC applications? > > Thanks, > Aleksey > From pierce at twinforces.com Wed Mar 17 08:32:03 1999 From: pierce at twinforces.com (Pierce T. Wetter III) Date: Thu Nov 3 14:41:53 2005 Subject: Swing In-Reply-To: <9903171611.AA01068@rkinc.com> Message-ID: At 8:27 AM -0800 3/17/99, Aleksey Sudakov wrote: > Hello, > > MacOSX Server comes with the JDK 1.1.6. It's quite outdated. DR2 also came > with JDK 1.1.6 which was dog slow and as a result could not be used to run > any Swing applications. Thus the question to anyone who have got MacOSX > server already: How good is Java performance and is there anybody who > successfully run any Swing/JFC applications? Java performance is 90% of native code after JITing. There shouldn't be any reason that you can't use swing with JDK 1.1.6, you just have to download swing.zip. I don't quite understand what you're looking for. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From zander at rkinc.com Wed Mar 17 08:51:06 1999 From: zander at rkinc.com (Aleksey Sudakov) Date: Thu Nov 3 14:41:53 2005 Subject: Swing References: Message-ID: <9903171651.AA01087@rkinc.com> : There shouldn't be any reason that you can't use swing with JDK : 1.1.6, you just have to download swing.zip. With JVM that came with RDR2 there was a huge reason - performance. JComponents did show up but that was pretty much it. JVM could not handle them sending events :) Aleksey From maury at OAAI.COM Wed Mar 17 09:09:40 1999 From: maury at OAAI.COM (Maury Markowitz) Date: Thu Nov 3 14:41:53 2005 Subject: OS4.2 builds and lib paths Message-ID: <199903171700.MAA06750@OAAI.COM> One of the _very_ nice upgrades to the OS system that Rhapsody included was that the later is a lot smarter about looking for libraries. This has simplified my build process quite a bit, and frankly just makes more sense than the OS4.2 side of things - what's the point of dynalibs if the compiler's going to hard link in the path? :-) Anyway the problem now is that I'm not at all sure how to go about removing all the dependencies from my OS4.2 builds. We use a rather nice build script on the DR2 side of things which builds into a temp directory and everything just works. When we moved that script back to the OS4.2 side of things everything _looked_ fine but then my beta testers noted it all pointed to my build directories (hard to test here, we only have one OS4.2 box and my dir is public). Grrr. I've looked in the makefile.postambles and found the DYLIB_INSTALL_DIR, but I'm not clear on what to do with it. Do I simply define it at the top of my script and that's that? I tried that but it didn't seem to do anything. Do I need to EXPORT it? Or do I need to put it into my linker flags? Do I need to change the postambles at all? Very confusing. Maury From jhendry at cmg.FCNBD.COM Wed Mar 17 09:25:47 1999 From: jhendry at cmg.FCNBD.COM (Jonathan Hendry) Date: Thu Nov 3 14:41:53 2005 Subject: [SUMMARY] Does MacOS X Server include an Oracle Adaptor In-Reply-To: <9903171624.AA08184@rednsi.com> References: <9903171624.AA08184@rednsi.com> Message-ID: <199903171725.LAA23879@wo1085.cmg.FCNBD.COM> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 540 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990317/64fc379e/attachment.bin From jjfeiler at diga.serv.net Wed Mar 17 09:53:40 1999 From: jjfeiler at diga.serv.net (John Jay Feiler) Date: Thu Nov 3 14:41:53 2005 Subject: MacOSX Server ships on the 23rd Message-ID: <9903171753.AA24151@relief.com> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 116 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990317/12893704/attachment.bin From kc at omnigroup.com Wed Mar 17 10:39:12 1999 From: kc at omnigroup.com (Ken Case) Date: Thu Nov 3 14:41:53 2005 Subject: Compatibility of Rhapsody > MacOSX-Server Message-ID: <199903171839.KAA22765@ignem.omnigroup.com> > Are Rhapsody binaries compatible with MacOSX Server (as long as > there is a PPC version) For the most part, Rhapsody DR2 binary applications will run without change on Mac OS X Server. However, there are a few incompatible changes cause problems with some applications. (For example, the SoundKit framework became a private framework, so any applications which linked against it will no longer launch because they can't find it. Fortunately, there's a simple workaround for this: "ln -s /System/Library/PrivateFrameworks/SoundKit.framework /Network/Library/Frameworks".) > and does Rhapsody Source Code compile out-of-the-box on MacOS-X > Server? Many predefined preprocessor symbols (like "NeXT" and "NS_TARGET_MAJOR") are gone, so any code which depends on those predefines (e.g. SystemType.h in OmniBase) no longer compiles without some minor porting effort (to look for __APPLE__ and __MACH__ and get the kernel version from instead). If you look back in the archives, you'll find a message by Wilfredo Sanchez which lists the predefined symbols which exist in Mac OS X Server. (To distinguish between changed API in various framework releases, we modified OmniBase to export OBOperatingSystemMajorVersion and OBOperatingSystemMinorVersion which the rest of our code now uses rather than NS_TARGET_MAJOR and NS_TARGET_MINOR. In an ideal world, each framework would export its own version number so we could look at which version of Foundation we were using, etc.) I intend to release updated source code for our Omni frameworks on our web site within the next week or two, since the currently published source code won't build on X Server. If anyone has an urgent need for this, let me know. Ken From tims at paradigmdev.com Wed Mar 17 10:42:56 1999 From: tims at paradigmdev.com (Tim Snider) Date: Thu Nov 3 14:41:54 2005 Subject: Swing Message-ID: <57B675B21506D1118BAB0060081C295D86A23B@vserver.paradigm.com> There is also the BlackDown port of Java 2.0, which while aimed at Linux could possibly be compiled/ported quickly. I know they've basically fixed the x86 version but still have a few outstanding bugs in the PPC version. Anyone have an idea as to how much work it would be to port it to MacOSX? Tim -----Original Message----- From: Pierce T. Wetter III [mailto:pierce@twinforces.com] Sent: Wednesday, March 17, 1999 9:54 AM To: Multiple recipients of list Subject: Re: Swing At 8:27 AM -0800 3/17/99, Aleksey Sudakov wrote: > Hello, > > MacOSX Server comes with the JDK 1.1.6. It's quite outdated. DR2 also came > with JDK 1.1.6 which was dog slow and as a result could not be used to run > any Swing applications. Thus the question to anyone who have got MacOSX > server already: How good is Java performance and is there anybody who > successfully run any Swing/JFC applications? Java performance is 90% of native code after JITing. There shouldn't be any reason that you can't use swing with JDK 1.1.6, you just have to download swing.zip. I don't quite understand what you're looking for. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From cwhite at comnetix.com Wed Mar 17 11:26:53 1999 From: cwhite at comnetix.com (Craig White) Date: Thu Nov 3 14:41:54 2005 Subject: Inter thread communication with NSConnection In-Reply-To: <199903170958.BAA20283@ignem.omnigroup.com> Message-ID: <9903171426.AA5397811@deli> I forgot to mention in my original post that I am calling 'setProtocolForProxy:' on the connection's rootProxy object. Craig From annard at webwarecorp.com Wed Mar 17 12:11:27 1999 From: annard at webwarecorp.com (Annard Brouwer) Date: Thu Nov 3 14:41:54 2005 Subject: expansion in PB on NT References: <9903171657.AA01404@slab> Message-ID: <36F00C67.FED15B63@webwarecorp.com> "Paul R. Summermatter" wrote: > I'm a long standing emacs user and am sorely > missing all of the emacs key bindings which I had on Mach. I have remapped > my keys so that I can use the left Control key as a true control, but none of > my escape sequences work (esc-d, esc-v, esc-shift-<, esc-shift->, and, in > particular, esc-/, which I use for expansion). If anyone has any suggestions > on how I can recover these bindings, especially a usable key binding for > expansion, I would appreciate hearing about it. > Hi people, Actually, extending Paul's request I would like to add the question of how you can remap the Control key to the Caps Lock key on almost all platforms (NT, Mac OS (8 and X Server). Being from the vt100 era I can't and will not get used to the impossible position of the Control key on "modern" platforms. Thanks, Annard P.S. Forwarded to MacOSX Dev... From bh40 at calva.net Wed Mar 17 12:43:25 1999 From: bh40 at calva.net (Benjamin Herrenschmidt) Date: Thu Nov 3 14:41:54 2005 Subject: Where are the mailing lists ? Message-ID: <19990317214325.006329@mail.mipsys.com> This is probably not the best place to ask this, but I'm a little bit clueless: there are some mailing lists referenced from the darwin site (after you login) but no infos about how to subscribe to them. Are developers automagically subscribed when registering ? -- E-Mail: BenH. Web : From pierce at alumni.caltech.edu Wed Mar 17 13:17:51 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:54 2005 Subject: expansion in PB on NT In-Reply-To: <36F00C67.FED15B63@webwarecorp.com> Message-ID: At 1:11 PM -0800 3/17/99, Annard Brouwer wrote: > "Paul R. Summermatter" wrote: >> I'm a long standing emacs user and am sorely >> missing all of the emacs key bindings which I had on Mach. I have remapped >> my keys so that I can use the left Control key as a true control, >> but none of >> my escape sequences work (esc-d, esc-v, esc-shift-<, esc-shift->, and, in >> particular, esc-/, which I use for expansion). If anyone has any >> suggestions >> on how I can recover these bindings, especially a usable key binding for >> expansion, I would appreciate hearing about it. >> > > Hi people, > > Actually, extending Paul's request I would like to add the question of > how you can remap the Control key to the Caps Lock key on almost all > platforms (NT, Mac OS (8 and X Server). Being from the vt100 era I can't > and will not get used to the impossible position of the Control key on > "modern" platforms. I think you can edit the keymap file in order to do this, but I'm not 100% sure. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From pierce at alumni.caltech.edu Wed Mar 17 13:25:01 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:54 2005 Subject: Where are the mailing lists ? In-Reply-To: <19990317214325.006329@mail.mipsys.com> Message-ID: At 1:23 PM -0800 3/17/99, Benjamin Herrenschmidt wrote: > This is probably not the best place to ask this, but I'm a little bit > clueless: there are some mailing lists referenced from the darwin site > (after you login) but no infos about how to subscribe to them. Are > developers automagically subscribed when registering ? Apple has a web-based signup at www.lists.apple.com. But the lists didn't exist the last time I checked. Pierce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From sanguish at digifix.com Wed Mar 17 13:34:47 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:54 2005 Subject: Compatibility of Rhapsody > MacOSX-Server In-Reply-To: <199903171319.OAA01493@xenia.hsh.stusta.mhn.de> References: <199903171319.OAA01493@xenia.hsh.stusta.mhn.de> Message-ID: <199903172134.QAA05281@digifix.com> Bernhard Scholz wrote: > Hello, > > Because we lack a G3 with MacOSX Server we currently can't figure > out what to do with our Rhapsody software on the Peanuts Archive > ( > We are currently thinking of including the Rhapsody stuff in a new > MacOSX Server directory or just to rename it. The later would imply > that compiled Rhapsody software is able to run out-of-the-box on > MacOS-X Server. Is this the case? > > Are Rhapsody binaries compatible with MacOSX Server (as long as there is > a PPC version) and does Rhapsody Source Code compile out-of-the-box > on MacOS-X Server? > Most stuff just runs. HOWEVER.. .anything that uses SoundKit will not startup... Its been replaced... From zander at rkinc.com Wed Mar 17 13:31:48 1999 From: zander at rkinc.com (Aleksey Sudakov) Date: Thu Nov 3 14:41:54 2005 Subject: Swing References: <57B675B21506D1118BAB0060081C295D86A23B@vserver.paradigm.com> Message-ID: <9903172131.AA00697@rkinc.com> : There is also the BlackDown port of Java 2.0, which while aimed at Linux : could possibly be compiled/ported quickly. : I know they've basically fixed the x86 version but still have a few : outstanding bugs in the PPC version. : : Anyone have an idea as to how much work it would be to port it to MacOSX? It's about the same as to port from Sun sources for Solaris and during last 3+ years nobody got full fledged Java on OpenStep however many tried Aleksey. From sanguish at digifix.com Wed Mar 17 13:38:10 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:54 2005 Subject: Changes in Yellow box... In-Reply-To: <199903171116.NAA18289@saulite> References: <199903171116.NAA18289@saulite> Message-ID: <199903172138.QAA05356@digifix.com> Nils Berzins wrote: > Hi, > > Now, when the Mac OS X Server is finally released (at least Macweek and > Stepwise claims so), it would be very nice if someone could put together > a FAQ (or something like that) about what's new and what's changed in > Yellow APIs since DR2. I suspect that it will take weeks before anyone > unfortunate enough to live outside the US will get their hands on Mac > OSXS... This was done a month and a half ago on Stepwise (thanks to YB 1.0 being in WOF 4.0). http://www.stepwise.com/Articles/Technical/GeekPorn.html From bbum at codefab.com Wed Mar 17 14:40:09 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:54 2005 Subject: Swing References: <9903172131.AA00697@rkinc.com> Message-ID: <36F02F49.F866CC08@mail.codefab.com> If somebody had a good rootless x server, the port of swing (or lots of other things) would be relatively moot. anyone looked at porting xfree86??? b.bum From pierce at alumni.caltech.edu Wed Mar 17 14:51:36 1999 From: pierce at alumni.caltech.edu (Pierce T. Wetter III) Date: Thu Nov 3 14:41:54 2005 Subject: Compatibility of Rhapsody > MacOSX-Server In-Reply-To: <199903172134.QAA05281@digifix.com> Message-ID: At 2:22 PM -0800 3/17/99, Scott Anguish wrote: > Bernhard Scholz wrote: >> Hello, >> >> Because we lack a G3 with MacOSX Server we currently can't figure >> out what to do with our Rhapsody software on the Peanuts Archive >> ( > >> We are currently thinking of including the Rhapsody stuff in a new >> MacOSX Server directory or just to rename it. The later would imply >> that compiled Rhapsody software is able to run out-of-the-box on >> MacOS-X Server. Is this the case? >> >> Are Rhapsody binaries compatible with MacOSX Server (as long as there is >> a PPC version) and does Rhapsody Source Code compile out-of-the-box >> on MacOS-X Server? >> > > Most stuff just runs. > > HOWEVER.. .anything that uses SoundKit will not startup... Its been > replaced... > Nope, moved. If you put in a symbolic link it will start up again. PIerce ---------------------------------------------------------------- Pierce T. Wetter III, Director, Twin Forces, Inc. e-mail: pierce@twinforces.com Phone:520-779-4227 U.S. Mail: 1300 South Milton Rd, Suite 206, Flagstaff, AZ 86001 From wsanchez at apple.com Wed Mar 17 15:08:47 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:54 2005 Subject: Where are the mailing lists ? Message-ID: <199903172308.PAA33950@scv1.apple.com> | This is probably not the best place to ask this, but I'm a little bit | clueless: there are some mailing lists referenced from the darwin site | (after you login) but no infos about how to subscribe to them. Are | developers automagically subscribed when registering ? Um... we're trying to fix that now. Kinda had too much to do yesterday. Missed a couple. :-) -Fred From afindlay at austin.rr.com Wed Mar 17 19:10:58 1999 From: afindlay at austin.rr.com (adhamh findlay) Date: Thu Nov 3 14:41:54 2005 Subject: Compatibility of Rhapsody > MacOSX-Server In-Reply-To: <199903172134.QAA05281@digifix.com> Message-ID: At 2:22 PM -0800 3/17/99, Scott Anguish wrote: >Bernhard Scholz wrote: >> Hello, >> >> Because we lack a G3 with MacOSX Server we currently can't figure >> out what to do with our Rhapsody software on the Peanuts Archive >> ( > >> We are currently thinking of including the Rhapsody stuff in a new >> MacOSX Server directory or just to rename it. The later would imply >> that compiled Rhapsody software is able to run out-of-the-box on >> MacOS-X Server. Is this the case? >> >> Are Rhapsody binaries compatible with MacOSX Server (as long as there is >> a PPC version) and does Rhapsody Source Code compile out-of-the-box >> on MacOS-X Server? >> > > Most stuff just runs. > > HOWEVER.. .anything that uses SoundKit will not startup... Its been >replaced... > > > Not replaced, moved. It is now in Private.frameworks. I have made applications work by placing a link in framework to soundkit in private.frameworks. That does the trick. adhamh From mvgfr at netcom.com Wed Mar 17 19:34:28 1999 From: mvgfr at netcom.com (Marc Farnum Rendino) Date: Thu Nov 3 14:41:54 2005 Subject: YellowBox/NT? In-Reply-To: <199903170154.TAA23077@wo1085.cmg.FCNBD.COM> Message-ID: At 6:00 PM -0800 on 99/03/16, Jonathan Hendry wrote: > > So- any word on YellowBox NT? > > Obviously, it's part of WebObjects. But can anyone *else* ship YB apps? You could ship a non-WO, YB app w/ a WO license... - Marc From td at twics.com Wed Mar 17 20:11:49 1999 From: td at twics.com (Troy Dawson) Date: Thu Nov 3 14:41:54 2005 Subject: Bluebox/OpenGL developments? In-Reply-To: Message-ID: I recall reading on Stepwise last summer that Conix had a beta of OpenGL for Rhapsody, but with the imminent shipping of MOSXS, I would like to know what the current state of OpenGL is now, with the various combinations of YB/BB and Conix/Mesa. thanks, =td= From ttakeo at amtec.co.jp Wed Mar 17 20:45:45 1999 From: ttakeo at amtec.co.jp (tetuya takeo) Date: Thu Nov 3 14:41:54 2005 Subject: International resource of MacOS X Server limited version Message-ID: <9903180437.AA26851@amtec000.amtec.co.jp> Hi, Anyone knows whether 5-client version of MacOS X Server (for ADC members) has international resources or not? In order to evaluate MOSXS as soon as possible, I'd like to order it. Thanks in advance. --- Tetuya TAKEO ttakeo@amtec.co.jp (NeXTmail/MIME accepted) From afindlay at austin.rr.com Wed Mar 17 21:15:12 1999 From: afindlay at austin.rr.com (adhamh findlay) Date: Thu Nov 3 14:41:54 2005 Subject: International resource of MacOS X Server limited version In-Reply-To: <9903180437.AA26851@amtec000.amtec.co.jp> Message-ID: I think it says on the site that these will be along shortly. Shouldn't take too long OpenStep was intenationalized. At 8:53 PM -0800 3/17/99, tetuya takeo wrote: >Hi, > >Anyone knows whether 5-client version of MacOS X Server (for ADC >members) has international resources or not? In order to evaluate >MOSXS as soon as possible, I'd like to order it. > >Thanks in advance. > >--- >Tetuya TAKEO >ttakeo@amtec.co.jp (NeXTmail/MIME accepted) adhamh From rmorse at rmsun.linxnet.com Thu Mar 18 04:03:15 1999 From: rmorse at rmsun.linxnet.com (Robert Morse) Date: Thu Nov 3 14:41:54 2005 Subject: ATI 128 Cards with MACOSX server Message-ID: <199903181203.HAA05250@rmsun.linxnet.com> Hi, It is great about MACOSXS finially comming out, but I do have one question which I cannot find an answer for. Now, It has been said that it should run, but unsupported on a 8600. Now, I really want one of those ATI Nexus 128 cards to put in it. Will MacOSXServer use it. I would think so, as it is almost the same as what comes in the B&W G3. Thanks Robert Morse ++++++++++++++++++++++++++++++++++++++++ + Robert Morse + + rmorse@rmsun.linxnet.com + ++++++++++++++++++++++++++++++++++++++++ From boerny at xenia.hsh.stusta.mhn.de Thu Mar 18 06:30:29 1999 From: boerny at xenia.hsh.stusta.mhn.de (Bernhard Scholz) Date: Thu Nov 3 14:41:54 2005 Subject: Darwin source availability In-Reply-To: <199903171839.KAA22765@ignem.omnigroup.com> References: <199903171839.KAA22765@ignem.omnigroup.com> Message-ID: <199903181430.PAA00757@xenia.hsh.stusta.mhn.de> I just got notified from European developers, that the Apple server is hopelessly overloaded. Therefore downloading the Darwin source from Apple is a torture. So the Peanuts-Archive decided to make the whole Darwin project available on the Peanuts-Archive at http://www.peanuts.org/ -> Mirrors/Apple/PublicSource/Darwin concerning to the APSL license paragraph 2.2 (Deployment of Covered Code). Of course you still have to accept the APSL before downloading :-) Greetings, Bernhard. --- Bernhard Scholz http://www.peanuts.org/ Peanuts-Archive Admin scholz@peanuts.org From jean-michel.cazaux at finindev.com Thu Mar 18 06:30:55 1999 From: jean-michel.cazaux at finindev.com (Jean-Michel Cazaux) Date: Thu Nov 3 14:41:54 2005 Subject: Pb on YB/95 without network adapter : please a quick help... Message-ID: <01BE7154.53CFEE40@JEAN-MICHEL> Hi folks, We are encountering a very big problem... We have to install a YB/EOF app on a W95 PC without network adapter... It's an EOF app, using an ODBC adaptor on an Interbase DB. The YB environement seems to work fine, when we launch the app it's correct too, we can read datas in the DB. *BUT*, as we try to insert datas, we got an error in WINSOCK.DLL ! ! ! We declared a TCP/IP protocol over a RAS adapter. We tried win 95 4.00.950 and 4.00.950B, the behaviour is the same. We use ODBC 3.0. Any help welcome, my boss is looking after me with it's riot gun ! ! ! Thanks in advance. __________________________________________________________________________________________ Jean-Michel CAZAUX FININDEV, Conseil en Finances pour les Collectivit?s Locales. 204 Rue Michel Teule - ZAC d'Alco 34080 Montpellier - FRANCE. T?l. +33 (0)4 67 63 66 25 - Fax +33 (0)4 67 63 35 45 e-mail: jean-michel.cazaux@finindev.com From sam at darth.geg.mot.com Thu Mar 18 10:19:01 1999 From: sam at darth.geg.mot.com (Sam M. Daniel) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions Message-ID: <9903181819.AA01391@darth.geg.mot.com> In anticipation of MOSXS, I have the following installation questions: 1. Just to verify - should I avoid installing MOSXS over RDR2? 2. If I install MOSXS on a different partition, should I format it to HFS+ first, or is that automatic? 3. With MOSXS running on its own partition, will I be able to see the RDR2 partition? Will I be able to see the MacOS partition? 4. Will I be able to trasfer files among the partitions? 5. Will I be able to print to an attached printer, specifically, an Epson Inkjet? More to the point, will there be support for popular non-postscript printers? Is there software that would allow a non-postscript printer to print postscript with MOSXS or MacOS? Or, will I need a Postscript printer, like the NeXT Laser or some other printer? 6. Will I be able to print from the Blue Box? 7. When MacOS 8.6 becomes available, will I be able to update the Blue Box from 8.5? Anxious to hear your response. Sam Daniel Motorola Inc. 602-441-7431 From pmb at apple.com Thu Mar 18 11:13:56 1999 From: pmb at apple.com (Peter Bierman) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: <9903181819.AA01391@darth.geg.mot.com> Message-ID: At 9:26 AM -0800 3/18/99, Sam M. Daniel wrote: >In anticipation of MOSXS, I have the following installation questions: > >1. Just to verify - should I avoid installing MOSXS over RDR2? "Upgrade" installations won't be possible. Installation *will* re-initialize the target volume. (Not the whole disk, just the volume.) >2. If I install MOSXS on a different partition, should I format it to HFS+ >first, or is that automatic? The installer will allow you to target an HFS, HFS+, or UFS volume for installation. You should do your disk partitioning before running the installer. (The installer won't install onto "free space". You need to allocate the space as an HFS or HFS+ volume. This volume will be erased and converted to UFS by the installer.) >3. With MOSXS running on its own partition, will I be able to see the RDR2 >partition? Will I be able to see the MacOS partition? I think so, and yes. >4. Will I be able to trasfer files among the partitions? Yes. >5. Will I be able to print to an attached printer, specifically, an Epson >Inkjet? More to the point, will there be support for popular non-postscript >printers? Is there software that would allow a non-postscript printer to >print postscript with MOSXS or MacOS? Or, will I need a Postscript printer, >like the NeXT Laser or some other printer? I don't know. >6. Will I be able to print from the Blue Box? Yes. >7. When MacOS 8.6 becomes available, will I be able to update the Blue Box >from 8.5? I don't know. -- "UNIX shells in Mac OS X should be unneeded but functional... and have the same installed base as MPW." From janos.lobb at yale.edu Thu Mar 18 11:39:55 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:54 2005 Subject: Fwd: MOSXS Installation and Related Questions Message-ID: One more question. Does MacOSXS support non-Apple SCSI in a PCI slot with drives attached to the card ? >Date: Thu, 18 Mar 1999 09:33:36 -0800 (GMT-0800) >Reply-To: sam@darth.geg.mot.com >Originator: macosx-dev@omnigroup.com >Sender: macosx-dev@omnigroup.com >Precedence: bulk >From: "Sam M. Daniel" >To: Multiple recipients of list >Subject: MOSXS Installation and Related Questions >X-Comment: To unsubscribe, follow directions at >http://www.omnigroup.com/MailArchive/ >Mime-Version: 1.0 (NeXT Mail 4.2mach v148) >Status: > >In anticipation of MOSXS, I have the following installation questions: > >1. Just to verify - should I avoid installing MOSXS over RDR2? >2. If I install MOSXS on a different partition, should I format it to HFS+ >first, or is that automatic? >3. With MOSXS running on its own partition, will I be able to see the RDR2 >partition? Will I be able to see the MacOS partition? >4. Will I be able to trasfer files among the partitions? >5. Will I be able to print to an attached printer, specifically, an Epson >Inkjet? More to the point, will there be support for popular non-postscript >printers? Is there software that would allow a non-postscript printer to >print postscript with MOSXS or MacOS? Or, will I need a Postscript printer, >like the NeXT Laser or some other printer? >6. Will I be able to print from the Blue Box? >7. When MacOS 8.6 becomes available, will I be able to update the Blue Box >from 8.5? > >Anxious to hear your response. > >Sam Daniel >Motorola Inc. >602-441-7431 > J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From pmb at apple.com Thu Mar 18 12:03:54 1999 From: pmb at apple.com (Peter Bierman) Date: Thu Nov 3 14:41:54 2005 Subject: Fwd: MOSXS Installation and Related Questions In-Reply-To: Message-ID: At 11:54 AM -0800 3/18/99, J?nos L?bb wrote: >One more question. Does MacOSXS support non-Apple SCSI in a PCI slot with >drives attached to the card ? Officially, no. Only a limited selection of machines and hardware is validated. That will depend on the exact card and firmware. You should ask Scott Anguish at www.stepwise.com to start setting up "compatibility grids" on web pages where people can send in their setup and whether it works or not. -pmb -- "UNIX shells in Mac OS X should be unneeded but functional... and have the same installed base as MPW." From smithja at cs.unc.edu Thu Mar 18 12:15:38 1999 From: smithja at cs.unc.edu (Jason M. Smith) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: Message-ID: >>5. Will I be able to print to an attached printer, specifically, an Epson >>Inkjet? More to the point, will there be support for popular non-postscript >>printers? Is there software that would allow a non-postscript printer to >>print postscript with MOSXS or MacOS? Or, will I need a Postscript printer, >>like the NeXT Laser or some other printer? > >I don't know. No serial support, and severly limited USB support. Unlikely, unless it's over Ethernet, and even then you'd need a proper driver. I've only tried Postscript over TCP/IP, and it worked beautifully. >>7. When MacOS 8.6 becomes available, will I be able to update the Blue Box >>from 8.5? > >I don't know. You should be able to... however: Easy way to ensure it - have your Blue Box boot volume be it's own physical volume. You can tell Blue Box (MacOS.app) which volume it should look to to boot from, the default is the pseudo-volume that come pre-installed. I have two drives in my machine, one MOSXS, the other MacOS 8.5.1. Launching Blue Box from within MOSXS boots up the *same* System Folder, etc, that I run when booting into vanilla MacOS 8.5.1. It's very convenient. If you want to update the Blue Box OS, just update that volume as you would normally. ------- -Jason From dev at humph.com Thu Mar 18 12:49:24 1999 From: dev at humph.com (Giuliano Gavazzi (dev)) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: Message-ID: At 8:37 pm +0000 1999/03/18, Jason M. Smith wrote: ... > No serial support, and severly limited USB support. Unlikely, I hope are you joking. Are we talking of MacOS X Server 1.0 here? What operating system is it if it has no serial support? I think we should transfer this to the admin list. Giuliano H U M P H || ||| software development of: Java & C++ Client/Human Interface applications. Native Server applications on MacOS - Rhapsody - (Mk)Linux http://www.humph.com/ From smithja at cs.unc.edu Thu Mar 18 13:01:21 1999 From: smithja at cs.unc.edu (Jason M. Smith) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: References: Message-ID: At 8:49 PM +0000 3/18/99, Giuliano Gavazzi (dev) wrote: >At 8:37 pm +0000 1999/03/18, Jason M. Smith wrote: >... >> No serial support, and severly limited USB support. Unlikely, > >I hope are you joking. Are we talking of MacOS X Server 1.0 here? What >operating system is it if it has no serial support? >I think we should transfer this to the admin list. It's only *supported* to run on the Blue & White G3s, remember... which have no serial ports. This isn't all that unexpected. Whether or not you can run in on a machine that has a true blue serial port is apparently your business. :/ Also, a USB -> serial converter apparently won't work, since the USB support is limited to *only* the keyboard and mouse, and even then just enough to get them working. (No hotswapping, for one thing.) Seems they're serious about dropping legacy hardware, and serial has been designated as such. It'd be nice if they'd provide full USB, as a replacement, but I'm not suprised it's not in 1.0... it isn't exactly a high priority for a server. ------- -Jason From Stephen.Lewallen at worldnet.att.net Thu Mar 18 13:22:00 1999 From: Stephen.Lewallen at worldnet.att.net (Stephen Lewallen) Date: Thu Nov 3 14:41:54 2005 Subject: Native library for MRJ compatible? Message-ID: <19990318211755.JOOR3249@[208.255.86.58]> Folks, If I build a native library (i.e. a library using JNI) on Mac OSX Server for the Mac OSX Server JDK 1.1.6, will this library be binary compatible to the MRJ on Mac OS 8.* ? Thanks! From dev at humph.com Thu Mar 18 13:53:13 1999 From: dev at humph.com (Giuliano Gavazzi (dev)) Date: Thu Nov 3 14:41:54 2005 Subject: serial support. Was: MOSXS Installation and Related Questions In-Reply-To: Message-ID: At 9:09 pm +0000 1999/03/18, Jason M. Smith wrote: >At 8:49 PM +0000 3/18/99, Giuliano Gavazzi (dev) wrote: >>At 8:37 pm +0000 1999/03/18, Jason M. Smith wrote: >>... >>> No serial support, and severly limited USB support. Unlikely, >> >>I hope are you joking. Are we talking of MacOS X Server 1.0 here? What >>operating system is it if it has no serial support? >>I think we should transfer this to the admin list. > > It's only *supported* to run on the Blue & White G3s, remember... >which have no serial ports. This isn't all that unexpected. > > Whether or not you can run in on a machine that has a true blue >serial port is apparently your business. :/ > > Also, a USB -> serial converter apparently won't work, since the >USB support is limited to *only* the keyboard and mouse, and even then just >enough to get them working. (No hotswapping, for one thing.) > > Seems they're serious about dropping legacy hardware, and serial >has been designated as such. It'd be nice if they'd provide full USB, as a >replacement, but I'm not suprised it's not in 1.0... it isn't exactly a >high priority for a server. You are half right, from the FAQ: Q. Does Mac OS X Server support all the Power Macintosh G3 build-to-order options? A. No. It supports ATA and SCSI drives, Ethernet, and limited USB (for keyboard and mouse). There is currently no support for serial ports, FireWire, or LocalTalk. And from the Installation guide: To use Mac OS X Server you must have a desktop or minitower Power Macintosh G3 computer with... So Desktop and Tower PMac G3s are supported, but serial nope. I would not call this legacy hardware, any G3 is all in all more modern than OS X server. How do they expect one to do PPP server or client anyway? Funnily enough Tom Vier has just done a port of darwin's serial driver to fix serial driver support on MkLinux. I guess this type of feedback is the main reason why Apple has opened the source. I have forwarded this to the admin list, if this thread is to continue shouldn't we drop it from the dev list? Giuliano From bbum at codefab.com Thu Mar 18 14:07:38 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:54 2005 Subject: serial support. Was: MOSXS Installation and Related Questions References: Message-ID: <36F1792A.1ED0FEDC@mail.codefab.com> (1) Serial Support If the darwin source has serial support, is it possible to replace the equivalent components in OS X Server with the darwin bits to gain serial support? (2) Is this appropriate for here? Yes and No. Yes because it is a MOSXS development issue-- i need occasional serial access as a developer (for my palm pilot framework, for example). No-- we really ought to have a darwin-dev mailing list.... and any questions aimed at goo below the foundation should probably be redirected to that list b.bum From sanguish at digifix.com Thu Mar 18 14:35:59 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:54 2005 Subject: Fwd: MOSXS Installation and Related Questions In-Reply-To: References: Message-ID: <199903182236.RAA15263@digifix.com> Peter Bierman wrote: > At 11:54 AM -0800 3/18/99, Janos Lobb wrote: > >One more question. Does MacOSXS support non-Apple SCSI in a PCI slot with > >drives attached to the card ? > > Officially, no. Only a limited selection of machines and hardware is > validated. > > That will depend on the exact card and firmware. You should ask Scott > Anguish at www.stepwise.com to start setting up "compatibility grids" on > web pages where people can send in their setup and whether it works or not. I've been working on this for a while now.. I'm looking for a little more third party information to add before rolling it out. If you successfully installed CR1 (not prerelease, not pirated stuff that has an unknown lineage.. please feed me the details of your system and the install at sanguish@digifix.com From rcfa at cubiculum.com Thu Mar 18 13:25:12 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: References: Message-ID: <199903182125.AA23954@kannix.cubiculum.com> What are the pros and cons of HFS+ vs. UFS for someone who has NO legacy of MacOS to support? How do the two compare in relevant features, speed (e.g. compilation), crash safety (i.e. automatic fsck or equivalent), etc? In other words, if I just do development, and some OSXS/OS-Mach/NS file serving, then what file system type is the recommended choice and why? Greetings, Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From bbum at codefab.com Thu Mar 18 14:57:15 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions References: <199903182125.AA23954@kannix.cubiculum.com> Message-ID: <36F184CB.920D2E7F@mail.codefab.com> Under RDR2, a system booted plain Mac OS (there are uses of plain old macos) couldn't see the UFS RDR volume. Considering that MacOS hasn't been patched, I suspect this is still true under Mac OS X Server. So, one advantage of HFS+ is that you can see the filesystem when booted into straight Mac OS. b.bum "Ronald C.F. Antony" wrote: > What are the pros and cons of HFS+ vs. UFS for someone who has NO > legacy of MacOS to support? How do the two compare in relevant > features, speed (e.g. compilation), crash safety (i.e. automatic > fsck or equivalent), etc? > > In other words, if I just do development, and some OSXS/OS-Mach/NS > file serving, then what file system type is the recommended choice > and why? > > Greetings, > > Ronald > ============================================================================== > "The reasonable man adapts himself to the world; the unreasonable one persists > in trying to adapt the world to himself. Therefore all progress depends on the > unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From sanguish at digifix.com Thu Mar 18 15:11:16 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: <199903182125.AA23954@kannix.cubiculum.com> References: <199903182125.AA23954@kannix.cubiculum.com> Message-ID: <199903182311.SAA15756@digifix.com> Ronald C.F. Antony wrote: > What are the pros and cons of HFS+ vs. UFS for someone who has NO > legacy of MacOS to support? How do the two compare in relevant > features, speed (e.g. compilation), crash safety (i.e. automatic > fsck or equivalent), etc? > > In other words, if I just do development, and some OSXS/OS-Mach/NS > file serving, then what file system type is the recommended choice > and why? UFS is the only supported FS for Unix work on OSXS. HFS+ is used by the Blue Box, AppleShare and NetBoot. You need no HFS+ partitions. From jayh at panix.com Thu Mar 18 14:37:46 1999 From: jayh at panix.com (Jay Hardesty) Date: Thu Nov 3 14:41:54 2005 Subject: Help linking to code compiled with the Portable Object Compiler Message-ID: <199903182237.RAA29881@panix2.panix.com> Hello, We have an application written in Yellow Box using the GNU cc compiler and are trying to figure how (or if it is even possible) to link in a library compiled using the Portable Object Compiler. Linking itself works, but it doesn't seem possible to use the objects compiled with the Portable Object Compiler (POC) from within code compiled with GNU cc. Passing plain C data back and forth (such as malloc'd C pointers) works fine, but attempting to sending objective C methods to an object returned from POC-compiled code generates a memory bus error. Explictly type-casting the returned object makes no difference. I assume that probably the two compilers have different ways of laying out the object data? Simply recompiling the POC-compiled library is not an option because the code relies heavily on the use of smalltalk-style blocks, which are only supported by the POC. I even tried recompiling the non-block portions of the POC-compiled library in GNU cc - just so that the objects with the appropriate name-mangling would exist on both sides of the compiler divide, but still just got memory bus errors. We're using version 2.2.14 of the Portable Object Compiler and Apple's GNU compiler version cc-771.4 (based on gcc version 2.7.2.1) under Rhapsody DR2. I hope this makes sense, and that it isn't a patently ridiculous question - I just don't know enough about compilers to know any better. Thanks very much for any pointers, Jay Hardesty jayh@panix.com From wsanchez at apple.com Thu Mar 18 17:32:15 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:54 2005 Subject: serial support. Was: MOSXS Installation and Related Questions Message-ID: <199903190132.RAA30446@scv1.apple.com> | Yes because it is a MOSXS development issue-- i need occasional serial access as a | developer (for my palm pilot framework, for example). | | No-- we really ought to have a darwin-dev mailing list.... and any questions | aimed | at goo below the foundation should probably be redirected to that list darwin-development@public.lists.apple.com is now up and is getting light traffic. Adding serial support to Darwin is a great idea, and you should be able to boot such a kernel on Mac OS X Server, assuming you don't break anything else in the process. -Fred From kcd at jumpgate.com Thu Mar 18 17:54:22 1999 From: kcd at jumpgate.com (Kenneth C. Dyke) Date: Thu Nov 3 14:41:54 2005 Subject: Help linking to code compiled with the Portable Object Compiler In-Reply-To: <199903182237.RAA29881@panix2.panix.com> References: <199903182237.RAA29881@panix2.panix.com> Message-ID: <199903190154.RAA16020@babylon5.jumpgate.com> Jay, Unfortunately you are probably completely hosed. There is no real standard for the object and runtime data for Objective C. Stepstone, POC, and GNU all have somewhat different layouts if I recall correctly (POC and Stepstone may be the same, it's been a while). The only way you could pontentially make your code work would be to write C wrappers around your POC code and include the POC runtime in your binary. It would suck, but ought to work. -Ken Jay Hardesty wrote: > Hello, > > We have an application written in Yellow Box using the GNU cc > compiler and are trying to figure how (or if it is even possible) to > link in a library compiled using the Portable Object Compiler. > Linking itself works, but it doesn't seem possible to use the objects > compiled with the Portable Object Compiler (POC) from within code > compiled with GNU cc. Passing plain C data back and forth (such > as malloc'd C pointers) works fine, but attempting to sending > objective C methods to an object returned from POC-compiled code > generates a memory bus error. Explictly type-casting the returned object > makes no difference. I assume that probably the two compilers have > different ways of laying out the object data? Simply recompiling the > POC-compiled library is not an option because the code relies heavily > on the use of smalltalk-style blocks, which are only supported by the > POC. I even tried recompiling the non-block portions of the > POC-compiled library in GNU cc - just so that the objects with the > appropriate name-mangling would exist on both sides > of the compiler divide, but still just got memory bus errors. > We're using version 2.2.14 of the Portable Object Compiler and Apple's > GNU compiler version cc-771.4 (based on gcc version 2.7.2.1) under > Rhapsody DR2. > I hope this makes sense, and that it isn't a patently ridiculous question > - I just don't know enough about compilers to know any better. > > Thanks very much for any pointers, > Jay Hardesty > jayh@panix.com > From infinity at ntwrks.com Thu Mar 18 18:14:56 1999 From: infinity at ntwrks.com (Sam Krishna) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: <199903182125.AA23954@kannix.cubiculum.com> References: <199903182125.AA23954@kannix.cubiculum.com> Message-ID: <9903190214.AA01060@ntwrks.com> Whilst expounding justly and with great righteousness about "Re: MOSXS Installation and Related Questions", on Thu, 18 Mar 1999 14:58:31 -0800 (GMT-0800) Anno Domini, "Ronald C.F. Antony" decreed: Ronald, It looks as if the UNIX file utilities (chmod, chown, etc) work on HFS+. MOSXS applies UNIX-style file permissions to HFS+ partitions when booted into it. You definitely want to keep cross-filesystem copying/moving to a minimum. It feels as if you're FTP'ing from a Windows-based FTP server when copying back-and-forth from HFS+-to-UFS or back again. Compilations are blindingly fast. It looks like about a factor-of-five difference from similarly clocked G3s vs. PIIs. (300 Mhz vs. 300 MHz). I've always thought building a large codebase was a better test of a system than Quake II or Unreal, IMHO. You'll be happy with your build speeds on a G3. :-) Sam > What are the pros and cons of HFS+ vs. UFS for someone who has NO > legacy of MacOS to support? How do the two compare in relevant > features, speed (e.g. compilation), crash safety (i.e. automatic > fsck or equivalent), etc? > > In other words, if I just do development, and some OSXS/OS-Mach/NS > file serving, then what file system type is the recommended choice > and why? > From theisen at akaMail.com Thu Mar 18 20:46:56 1999 From: theisen at akaMail.com (Dirk Theisen) Date: Thu Nov 3 14:41:54 2005 Subject: Darwin Lists up! In-Reply-To: <36F1792A.1ED0FEDC@mail.codefab.com> Message-ID: <1dow162.dn39wcv7fdeM@ascend-tk-p23.rhrz.uni-bonn.de> bbum@codefab.com (Bill Bumgarner): > No-- we really ought to have a darwin-dev mailing list.... and any > questions aimed at goo below the foundation should probably be redirected > to that list They are up now: ------------------------------------------------- Welcome to the darwin-development mailing list! Please save this message for future reference. Thank you. If you ever want to remove yourself from this mailing list, send the following command in email to : ------------------------------------------------- Dirk -- Buy a Pentium III now, get your personal Big Brother FREE!!! http://theisen.home.pages.de/ From scott_ribe at killerbytes.com Thu Mar 18 21:05:20 1999 From: scott_ribe at killerbytes.com (Scott Ribe) Date: Thu Nov 3 14:41:54 2005 Subject: serial ports Message-ID: <19990318230520.014256@smtp.scott.net> OK, I see in the Apple info that "serial ports are not supported" or some such. But what does this really mean? I have an app that would be a good WebObjects pilot project for me. But it requires access to the serial ports. Not any kind of standard driver or abstraction that is expected by any particular application software. Just the ability to, from C code, open, send, and receive. Can that be done? Or is support of serial ports *completely* lacking? Scott Ribe scott_ribe@killerbytes.com http://www.scott.net/~sribe (205) 591-9204 voice (205) 591-8167 fax From ttakeo at amtec.co.jp Thu Mar 18 21:39:44 1999 From: ttakeo at amtec.co.jp (tetuya takeo) Date: Thu Nov 3 14:41:54 2005 Subject: serial ports In-Reply-To: <19990318230520.014256@smtp.scott.net> References: <19990318230520.014256@smtp.scott.net> Message-ID: <9903190531.AA28682@amtec000.amtec.co.jp> Hi, > Not any kind of standard driver or abstraction that is expected by > any particular application software. Just the ability to, from C code, > open, send, and receive. Can that be done? Or is support of serial ports > *completely* lacking? > In other words, are some driver frameworks for MacOS X Server released for developers? Or, should wait until release of IOKit? --- Tetuya TAKEO ttakeo@amtec.co.jp (NeXTmail/MIME accepted) From rcfa at cubiculum.com Thu Mar 18 21:39:40 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: <9903190214.AA01060@ntwrks.com> References: <9903190214.AA01060@ntwrks.com> Message-ID: <199903190539.AA26439@kannix.cubiculum.com> > Compilations are blindingly fast. It looks like about a factor-of-five > difference from similarly clocked G3s vs. PIIs. Are these on a UFS or HFS+ partition, or both? Just trying to get a feel of the performance curve of the two file system types... Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From ICI at questintl.com Fri Mar 19 00:52:01 1999 From: ICI at questintl.com (ICI) Date: Thu Nov 3 14:41:54 2005 Subject: serial ports Message-ID: <001021BA.004226@questintl.com> Message re-sent by ICI Administrator due to earlier cc:Mail problem. ______________________________ Reply Separator _________________________________ Subject: serial ports Author: Scott Ribe at Internet Date: 3/18/99 9:10 PM OK, I see in the Apple info that "serial ports are not supported" or some such. But what does this really mean? I have an app that would be a good WebObjects pilot project for me. But it requires access to the serial ports. Not any kind of standard driver or abstraction that is expected by any particular application software. Just the ability to, from C code, open, send, and receive. Can that be done? Or is support of serial ports *completely* lacking? Scott Ribe scott_ribe@killerbytes.com http://www.scott.net/~sribe (205) 591-9204 voice (205) 591-8167 fax From discord at charm.net Fri Mar 19 06:16:51 1999 From: discord at charm.net (dIsCoRd) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: <199903190539.AA26439@kannix.cubiculum.com> Message-ID: On Thu, 18 Mar 1999, Ronald C.F. Antony wrote: > Are these on a UFS or HFS+ partition, or both? Just trying to get a feel Alright. I'm confused. The install guide seems to imply that you can only install MXS on a UFS partition. You can have lots of HFS+ partitions lying around that MXS can use, but the install has to be on a UFS partition. Is that true? Or can you install on a HFS+ partition entirely? -- /\ karl hsu / \ discord@charm.net / \ www.charm.net/~discord/ / () \ / \ / \____/ \ /____________\ From mike at lorax.com Fri Mar 19 07:37:15 1999 From: mike at lorax.com (Mike Ferris) Date: Thu Nov 3 14:41:54 2005 Subject: expansion in PB on NT Message-ID: <199903191632.IAA24338@boom.lorax.com> > > I'm a long standing emacs user and am sorely > > missing all of the emacs key bindings which I had on Mach. I have remapped > > my keys so that I can use the left Control key as a true control, but none of > > my escape sequences work (esc-d, esc-v, esc-shift-<, esc-shift->, and, in > > particular, esc-/, which I use for expansion). If anyone has any suggestions > > on how I can recover these bindings, especially a usable key binding for > > expansion, I would appreciate hearing about it. You can customize the key bindings in the text system. Key bindings can include multi-key sequences such as esc-d, etc... See the Developer/YellowBox/TasksAndConcepts/ProgrammingTopics documentation. There's a document called TextDefaultsAndBindings which describes all this stuff and more. Mike Ferris From rcfa at cubiculum.com Fri Mar 19 10:09:54 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:54 2005 Subject: MOSXS Installation and Related Questions In-Reply-To: References: Message-ID: <199903191810.AA29611@kannix.cubiculum.com> > > >Compilations are blindingly fast. It looks like about a factor-of-five > > >difference from similarly clocked G3s vs. PIIs. > > > > Are these on a UFS or HFS+ partition, or both? Just trying to get a feel > > Alright. I'm confused. The install guide seems to imply that you can only > install MXS on a UFS partition. You can have lots of HFS+ partitions lying > around that MXS can use, but the install has to be on a UFS partition. > Is that true? Or can you install on a HFS+ partition entirely? It obviously seems to be the case that the HFS+ as primary file system rumor is not true, at least not for OSXS. That might change for OSX. However, that doesn't answer the question of relative performance. Since OSXS can access HFS+, it is well possible to e.g. put a source tree on an HFS+ volume and to compile it there. So the question remains: were these compilations done on a UFS or HFS+ volume, or does the fs type not have much of an influence on compiler performance in the case of OSXS? Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From fujisawa at the.canon.co.jp Fri Mar 19 18:28:48 1999 From: fujisawa at the.canon.co.jp (Jun Fujisawa) Date: Thu Nov 3 14:41:54 2005 Subject: Swing In-Reply-To: Message-ID: At 10:02 AM -0800 99.3.17, Pierce T. Wetter III wrote: > There shouldn't be any reason that you can't use swing with JDK > 1.1.6, you just have to download swing.zip. I don't quite understand > what you're looking for. The latest release - Swing 1.1.1 beta 1 works only with the JDK 1.1.7. or later. I've heard it's because JDK 1.1.6 has some significant bugs that affect focus in lightweight components. -- Jun Fujisawa From paulrs at lgs-systems.com Sat Mar 20 06:35:39 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:54 2005 Subject: What is going on with NSPopUpButton Message-ID: <9903201435.AA04836@slab> RE YB/NT, WOF 4.0 Hey guys, It appears that the NSPopUpButton got a complete rewrite since OpenStep 4.2, and I don't care for the results so far. If anyone can confirm these problems on MacOS X Server and/or provide any suggestions for workarounds, I would greatly appreciate it. I've read the release notes for the AppKit and documentation on NSPopUpButton, and, though this information confirms major changes, I don't see any information that would lead me to believe that our existing code/nibs should fail. Out first problem is that there is no visual cue when a pop up button is the first responder. On OS 4.2, a dotted outline would be drawn to show first responder status. NSButtons still do this, but not NSPopUpButtons. Why? Do we need to do a poseAs and start drawing on our own, or will this be fixed soon? Secondly, there is a nasty bug involving using the keyboard to navigate the items in the pop-up list when the list is selected on an empty title. (see the example at http://www.lgs-systems.com/Examples/PopUpCrashEOF.tgz). Thirdly, if you successfully navigate to a pop up using the keyboard and you hit the space bar to pop-up the list, if the list extends beyond the bottom of the screen and you use the keyboard to arrow down, the list does not move up, as it does when you use the mouse. (In all fairness, this problem existed in OS 4.2, but we were hoping it would be fixed.) Finally, when you configure an NSPopUpButton to pull down rather than pop up, how do you go about creating a title for the button. This code, which worked perfectly well on OS 4.2 fails to work on YB/NT: _jumpPullDown = [[[NSPopUpButton alloc] init] autorelease]; [_jumpPullDown setTarget:self]; [_jumpPullDown setAction:@selector(jump:)]; [_jumpPullDown setPullsDown:YES]; [_jumpPullDown setTitle:LGS_JUMP]; [_jumpPullDown ... // add other items with titles The pull down that results has no item named LGS_JUMP, but has all of the other items which were added using addItemWithTitle. Also, I can't seem to configure this from IB. Once you have dragged a pop up button off the palette and set it to be a pull down, there seems to be no way to set the title. Interestingly, our existing OS 4.2 nibs with pull downs, when opened in IB, show no titles, but when run in our app do have a title. How does that work? If anyone can shed some light on these issues or point me to documentation which explains all of the changes to NSPopUpButton, I would greatly appreciate it. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From paulrs at lgs-systems.com Sat Mar 20 11:16:54 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:54 2005 Subject: Addendum to NSPopUpButton Problems - MORE PROBLEMS Message-ID: <9903201916.AA05024@slab> Hey guys, I found one other issue which has changed regarding pop up buttons. If you have a pop up button with its action set and in the action method you do something like: - (void)foo:(id)sender { int tag = [[sender selectedCell] tag]; ... } you don't get what you used to (that is, you don't get what you did on OS 4.2 which was the item that was chosen from the list), instead, you get the NSPopUpButtonCell. I'm not arguing that this doesn't make sense, I just can't seem to find this change in API documented! Also, in trying to find a workaround for my problem of being unable to set a title for a pull down button in IB, I decided to try the following. Drag an NSPopUpButton off the palette. Before switching to a pull down, change 'Item 1' to 'THE TITLE'. Then, switch the button to a pull down. If your system behaves as mine does, 'THE TITLE', gets chopped to 'THE TIT'. Nice, huh! Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From paulrs at lgs-systems.com Sat Mar 20 17:14:45 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:54 2005 Subject: NSPopUpButton - Last (hopefully) Message-ID: <9903210114.AA05170@slab> Hey guys, After finding the MenuMadness example and looking it over, I think I was a bit unfair in my first post on this topic. Though I am frustrated by some of the changes and the lack of support in IB, it looks like Mike incorporated some wonderful new features into NSPopUpButton. I would still like to resolve the issues about which I posted, but at least, in the interim, there is an upside to this. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From yami at dementian.com Sun Mar 21 05:54:17 1999 From: yami at dementian.com (Steven D. Arnold) Date: Thu Nov 3 14:41:54 2005 Subject: real-time scheduling? Message-ID: <199903211435.JAA29055@phear.dementian.com> Now that the Mach kernel is open-sourced, wouldn't it be cool to add real-time scheduling capabilities to the kernel (or something close to real-time scheduling)? We can certainly add something like this: 1. Process provides a deadline for an action to occur. 2. As we approach the deadline, the system automatically increases the priority of the process, until all available resources are devoted to it. (We might specify maximum resources for different kinds of processes, different users, etc...) 3. If different real-time processes were competing for resources, we'd need some way to resolve the competition. This kind of capability could prove very important for both workstations and things like computers that write CDs. Workstations are important because an end-user expects rapid response. So the frontmost application should receive a high real-time priority. Just thoughts... steve From penrose at sfc.keio.ac.jp Sun Mar 21 06:14:31 1999 From: penrose at sfc.keio.ac.jp (Christopher Penrose) Date: Thu Nov 3 14:41:54 2005 Subject: real-time scheduling? References: <199903211435.JAA29055@phear.dementian.com> Message-ID: <36F4FEC2.306850E0@sfc.keio.ac.jp> "Steven D. Arnold" wrote: > > Now that the Mach kernel is open-sourced, wouldn't it be cool to add > real-time scheduling capabilities to the kernel (or something close to > real-time scheduling)? > > We can certainly add something like this: > > 1. Process provides a deadline for an action to occur. > 2. As we approach the deadline, the system automatically increases the > priority of the process, until all available resources are devoted to it. > (We might specify maximum resources for different kinds of processes, > different users, etc...) > 3. If different real-time processes were competing for resources, we'd need > some way to resolve the competition. > > This kind of capability could prove very important for both workstations and > things like computers that write CDs. Workstations are important because an > end-user expects rapid response. So the frontmost application should > receive a high real-time priority. The only drawback that I can think of, and a big one in RTMach, is that the overhead costs of real-time scheduling slow the kernel down tremendously, as well as increase the time latency in which an action can be completed by the kernel. The RTMach lab here at Keio has come to that consensus; its a tragic paradox. It might be more efficient and simple to hone and use the existing priority framework. Christopher penrose@sfc.keio.ac.jp From penrose at sfc.keio.ac.jp Sun Mar 21 06:28:47 1999 From: penrose at sfc.keio.ac.jp (Christopher Penrose) Date: Thu Nov 3 14:41:54 2005 Subject: real-time scheduling? References: <36F4FEC2.306850E0@sfc.keio.ac.jp> Message-ID: <36F50217.64E605@sfc.keio.ac.jp> Christopher Penrose wrote: > It might be more efficient and simple to hone and use the existing priority > framework. I meant to say more: use the existing priority framework from user space, rather than burden the kernel. Christopher penrose@sfc.keio.ac.jp From jdevlin at blacklab.la.asu.edu Sun Mar 21 15:48:16 1999 From: jdevlin at blacklab.la.asu.edu (John Devlin) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" Message-ID: <9903212348.AA00857@blacklab.la.asu.edu> This is a long, and probably even more naive. I'm just trying to see if I understand the concepts involved; and the only way to test my understanding is to see if I can apply it to a concrete situation. Any help would be appreciated. Here goes. There seems to be a longstanding debate about the merits of directly accessing instance variables. One school of thought says that you should write accessor methods for every instance variable and avoid direct access. The benefit is supposed to be more flexible code, since the other methods in the class are not making any assumptions about instance variables. For example, if a class named "Point" has instance variables named "x-co-ordinate" and "y-co-ordinate" and you later decide to work with polar co-ordinates instead, you only have to change the accessor methods; the rest of the methods are insulated from the underlying representation of the data. The other school of thought says that direct access makes code more efficient (since you avoid a message send every time you get or set an instance variable) and easier to read. But the main complaint seems to be that writing accessor methods for every instance variable opens up the object to inspection and manipulation which undermines the encapsulation of its data. I take it that the kinds of trade-offs involved here are well understood, even if there is no consensus on what weight to give these competing considerations. I only know what I read, and I'm relying here on bits and pieces from Beck, _Smalltalk Best Practice Patterns_ Skublics et. al., _Smalltalk With Style_ Chamond Liu, _Smalltalk, Objects and Design_ along with a paper of Beck's (with the polar co-ordinates example) and some posts to this group under the thread "Programming Style Issues". So if I'm wrong on any of these assumptions, please let me know. (Aside: Beck's book gives careful consideration to both sides, Skublics et.al. recommend accessors without reservation, and Liu has a brief overview of the debate. Liu's book is in general the most helpful book I've read on OOP, but this is from the persepective of someone who needed a lot of help.) Now here's the concrete part. For the most part, I've only seen this issue discussed in relation to Smalltalk. I'm wondering whether DIFFERENT issues arise in relation to Objective-C. For example, consider the implementation of the Country class in the TravelAdvisor.app from the Developer Tutorial. EVEN IF one was inclined to avoid directly accessing variables, should one avoid direct access even when implementing the -init and -dealloc methods? Suppose you decide to eliminate direct access. In the implementation of -init, you'd have a line something like [self setName:@""]; as opposed to name = @""; But if the -setName:newName method is implemented as follows: NSString *oldName = name; name = [newName copy]; [oldName release]; you'd be using "name" before it's initialized in the first line and in that case I don't know WHAT to make of the third line. So I'm wondering whether -init and -dealloc methods should be a special case EVEN IF one favors accessors over direct access. And are there other sorts of methods that should be thought of as special cases? ---- It's also hard to see how best to implement -initWithCoder: without using direct access. The implementation found in the Tutorial includes the line [coder decodeValueOfObjCType:@encode(float) at:¤cyRate]; (this is from memory, but I hope you get the idea). The only thing I can think of is to write a method (float *)currencyRateReference { return ¤cyRate; } and then rewrite the line above as [coder decodeValueOfObjCType:@encode(float) at:[self currencyRateReference]]; or is there a better solution? ---- On the positive side, it seems as if using accessors would facilitate memory management since the job of releasing old values and retaining or copying new ones would be the responsibility of the setter methods. For example, when implementing -initWithCoder instead of having worrying about copying the decoded object as in name = [[coder decodeObject] copy]; you'd just write [self setName[coder decodeObject]]; and let -setName:newName worry about copying it. Is this advantage real or illusory? Are there other advantages to using accessors in Objective-C that would not be apparent from a discussion of the issue as it arises in Smalltalk? For example, would it make distribution easier to retrofit into an app? Thanks in advance for your insight. Jd. From dyoung at vviuh221.vvi.com Sun Mar 21 17:06:57 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" References: <9903212348.AA00857@blacklab.la.asu.edu> Message-ID: <9903220106.AA03363@vviuh221.vvi.com> > There seems to be a longstanding debate I scanned your e-mail and I didn't see the "fragile base class" argument in your descriptions. That is one of the main reasons for using set/get methods for super-class ivars. Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From anders at milkweed.com Sun Mar 21 17:13:51 1999 From: anders at milkweed.com (Anders Pytte) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <9903212348.AA00857@blacklab.la.asu.edu> Message-ID: >This is a long, and probably even more naive. I'm just trying to see if >I understand the concepts involved; and the only way to test my >understanding is to see if I can apply it to a concrete situation. Any >help would be appreciated. Here goes. I'm sure I am atleast as naive as you, but I am eager to respond all the same. >There seems to be a longstanding debate about the merits of directly >accessing instance variables. One school of thought says that you should >write accessor methods for every instance variable and avoid direct >access. The benefit is supposed to be more flexible code, since the >other methods in the class are not making any assumptions about instance >variables. For example, if a class named "Point" has instance variables >named "x-co-ordinate" and "y-co-ordinate" and you later decide to work >with polar co-ordinates instead, you only have to change the accessor >methods; the rest of the methods are insulated from the underlying >representation of the data. I think the advantages of private data are overwhelming where efficiency is not an issue. >The other school of thought says that direct access makes code more >efficient (since you avoid a message send every time you get or set an >instance variable) and easier to read. But the main complaint seems to >be that writing accessor methods for every instance variable opens up the >object to inspection and manipulation which undermines the encapsulation >of its data. Why is an object with private data and public accessors more open to inspection an object with public data? Don't write accessors for data that you don't want public. In a fully dynamic language such as ObjC the compiler is never able to inline accessors, so efficiency is a factor in that case. In my opinion, Eiffel comes closest to ideal in data hiding. An object's members are by default visible but not mutable. That is, the variable name is also its accessor, which one can extend if one wants more complex behavior, without requiring any change to clients. Mutation requires always an explicit Set method. Since the language performs static optimizations, most accessors can be inlined for efficiency. [snip] >Now here's the concrete part. For the most part, I've only seen this >issue discussed in relation to Smalltalk. I'm wondering whether >DIFFERENT issues arise in relation to Objective-C. > >For example, consider the implementation of the Country class in the >TravelAdvisor.app from the Developer Tutorial. EVEN IF one was inclined >to avoid directly accessing variables, should one avoid direct access >even when implementing the -init and -dealloc methods? Suppose you >decide to eliminate direct access. In the implementation of -init, you'd >have a line something like > > [self setName:@""]; > >as opposed to > > name = @""; > >But if the -setName:newName method is implemented as follows: > > NSString *oldName = name; > name = [newName copy]; > [oldName release]; > >you'd be using "name" before it's initialized in the first line and in >that case I don't know WHAT to make of the third line. The -init method is essentially a constructor, that is, it is responsible for putting the object into a valid (self consistent) state. Since methods of self are not normally guaranteed for an object that is not in a self consistent state, one would normally try to avoid calling methods defined at that level of self in -init. One should call superclass methods on self after [super init], however. Of course, we all break that rule, but certainly it is normal NOT to call field accessors from -init just as one never calls a field accessor from itself (creating an infinite loop). The great advantage of using accessors even within the scope of a class is that then a derived class can override the accessor to get some desired side effect. In that case, the constructor (or -init method) for the derived class must ensure the desired effect of the Set method without relying on Set being called. [snip] I found the rest of your discussion reasonable though not particularly compelling. Regards, Anders. ---------------------------------------------------------------- Anders Pytte Milkweed Software Ferguson Hill Voice: (802) 472-5142 Cabot VT 05647 Internet: anders@milkweed.com ---------------------------------------------------------------- From cdoherty at skidmore.edu Sun Mar 21 21:33:19 1999 From: cdoherty at skidmore.edu (Fuzzy Logic) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <199903220509.VAA27555@ignem.omnigroup.com> Message-ID: >even when implementing the -init and -dealloc methods? Suppose you >decide to eliminate direct access. In the implementation of -init, you'd >have a line something like > > [self setName:@""]; > >as opposed to > > name = @""; I don't see any sense in restricting direct access within the object. accessors are there as part of the external interface, so the implementation can change (say from simply returning a variable to a more complex function). the purpose of encapsulation is to hide stuff from the *outside*. that's one of my general rules I use for OO design, one of the things for which it was designed: keep the interface consistent, and use the features that are there for that purpose. to that end I've been known to put in stupid-looking accessor functions, just out of principle. mind you, I come mainly from Java, where things are much more static than Obj-C; I had forgotten than Obj-C can't necessarily optimize by inlining accessors. still, I'm inclined to take a slight performance hit if it makes code more maintainable, since I hate nothing so much as having to find all the instances of something (say direct access) in a program, and change them. :-) Chris cdoherty@skidmore.edu ------------------------------- Chris Doherty cdoherty@skidmore.edu If you cannot find the truth right where you are, where else do you expect to find it? -- Dogen It is from moments of confusion that we develop moments of clarity. From cdoherty at skidmore.edu Sun Mar 21 22:33:03 1999 From: cdoherty at skidmore.edu (Fuzzy Logic) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: References: Message-ID: [many good points snipped] >I'm beginning to feel more strongly that always using the set method >should be a requirement in my code... not so strongly that I'm making >noise when my colleagues don't follow it, but always using set methods has >saved me *lots* of maintenance coding time. Actually, let me rephrase >that-- NOT always using the set methods has cost me enough tedious hours >of undoing various direct assignments to make me wish that I had always >followed the convention. hmm...I'm starting to think that a broadside policy stops once you get to the object internals, at which point you have to decide which will work better...I would base it on how likely it is that I would be dealing with subclasses, and just how different those subclasses might be. since the subclasses will have the variables anyway, you'll have to overload the set methods...I guess it's a question of whether you want to be bothered with overloading init and alloc and their friends as well when the implementation changes. I can easily envision (and have dealt with) both cases, so I guess I can't come down one way or the other. :-) Chris ------------------------------- Chris Doherty cdoherty@skidmore.edu If you cannot find the truth right where you are, where else do you expect to find it? -- Dogen It is from moments of confusion that we develop moments of clarity. From bbum at codefab.com Sun Mar 21 22:24:50 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:54 2005 Subject: Scripted ObjC and WebObjects/Yellow Box. Message-ID: Scott Anguish was kind enough to publish a write-up of the MultiScript Shell that I have been working on. http://www.stepwise.com/Articles/Technical/codefab/mssh1.html I wrote the article both as an introduction to mssh (and its ability to execute WebScript at the command line like any other shell script) as well as a glimpse at the inner workings of Apple's rather amazing MultiScript (the interpretive engine behind WebScript). For WebObjects folks, this may give you some ideas as to how to more effectively write WebScript code that is easier to read and easier to debug. For Yellow Box Folks, it provides a means of accessing a truly awesome, cross platform (including Yellow on NT as long as you have WO installed), scripting solution that is fully compatible with the AppKit and all other frameworks on the system. b.bum From bbum at codefab.com Sun Mar 21 22:19:48 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: Message-ID: Ya'll knew I could resist hopping in on this conversation.... :-) On Sun, 21 Mar 1999, Fuzzy Logic wrote: > >even when implementing the -init and -dealloc methods? Suppose you > >decide to eliminate direct access. In the implementation of -init, you'd > >have a line something like > > > > [self setName:@""]; > > > >as opposed to > > > > name = @""; > > I don't see any sense in restricting direct access within the object. > accessors are there as part of the external interface, so the > implementation can change (say from simply returning a variable to a more > complex function). the purpose of encapsulation is to hide stuff from the > *outside*. Except when you are subclassing the object and want to constrain the variable initialization or setting to some set of values. By making sure everything-- including the designated initializer (which may change or bifurcate between super and subclasses)-- goes through the set method, it means that any changes to support variable constraint or other value maintenance issues need only be coded in one place. As well, if the variable's type, meaning or function were ever to change-- including potentially splitting the variable across multiple other variables-- following the rule of "everything must use the set method", again, leads to easier maintenance with less potential risk of missing various edge cases. I'm beginning to feel more strongly that always using the set method should be a requirement in my code... not so strongly that I'm making noise when my colleagues don't follow it, but always using set methods has saved me *lots* of maintenance coding time. Actually, let me rephrase that-- NOT always using the set methods has cost me enough tedious hours of undoing various direct assignments to make me wish that I had always followed the convention. > that's one of my general rules I use for OO design, one of the things for > which it was designed: keep the interface consistent, and use the features > that are there for that purpose. to that end I've been known to put in > stupid-looking accessor functions, just out of principle. It seems that in the interest of keeping the interface consistent, one should generally try to use only one of the various primitives in the interface for doing any given task. Direct assignment and the set methods are both available primitives; by choosing to use one and only one-- the set method-- it makes for more consistent use of the interface. Of course, one could also say that it would be use to always use -valueForKey:. However, the key/value coding stuff loses in compiled code in that it eliminates all benefits of static type checking (thanks, christian, i was listening :-). > mind you, I come mainly from Java, where things are much more static than > Obj-C; I had forgotten than Obj-C can't necessarily optimize by inlining > accessors. still, I'm inclined to take a slight performance hit if it makes > code more maintainable, since I hate nothing so much as having to find all > the instances of something (say direct access) in a program, and change > them. :-) Agreed. Totally and completely. Given the vast improvement in processor speed and memory, the cost benefits of trying to optimize dynamic code by making it more static are not nearly as attractive when those optimizations also increase maintenance cost and time. b.bum From smithja at cs.unc.edu Sun Mar 21 22:56:39 1999 From: smithja at cs.unc.edu (Jason M. Smith) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: Message-ID: At 9:45 PM -0800 3/21/99, Fuzzy Logic wrote: >>even when implementing the -init and -dealloc methods? Suppose you >>decide to eliminate direct access. In the implementation of -init, you'd >>have a line something like >> >> [self setName:@""]; >> >>as opposed to >> >> name = @""; > >I don't see any sense in restricting direct access within the object. >accessors are there as part of the external interface, so the >implementation can change (say from simply returning a variable to a more >complex function). the purpose of encapsulation is to hide stuff from the >*outside*. A slight clarification here, IMHO... the purpose of encapsulation is to hide non-essential details from *the developer*... that may include the developer(s) of the original class, and definitely includes those developing subsequent inherited classes. Generally, most people consider 'the developer' to just be anyone external who uses that class, forgetting that the original creator, as well as anyone extending the class, are also included in this. YMMV, of course. >that's one of my general rules I use for OO design, one of the things for >which it was designed: keep the interface consistent, and use the features >that are there for that purpose. to that end I've been known to put in >stupid-looking accessor functions, just out of principle. The important thing to remember about guidelines is that you should know when to break them. Within -init and -dealloc methods is one place where you may need to *bypass* the accessors, even if they are available to the external interface. >mind you, I come mainly from Java, where things are much more static than >Obj-C; I had forgotten than Obj-C can't necessarily optimize by inlining >accessors. still, I'm inclined to take a slight performance hit if it makes >code more maintainable, since I hate nothing so much as having to find all >the instances of something (say direct access) in a program, and change >them. :-) Agreement here - I generally use the accessor methods by default for maintenance and development, then revert to a direct access for optimization only... you know, the way you're *supposed* to optimize. Implement, test, prioritize, optimize. :) In a system like C++ or Java, simple accessors don't make much of an overall performance hit, and I can usually leave them in within the implementation. Obj-C is something a bit different, and if you find yourself replacing many (or most) of these accessor calls, then perhaps they should be unused internally by default. But *please*, unless absolutely necessary for speed, keep them in the external interface. ------- -Jason From rcfa at cubiculum.com Sun Mar 21 23:50:17 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <9903212348.AA00857@blacklab.la.asu.edu> References: <9903212348.AA00857@blacklab.la.asu.edu> Message-ID: <199903220750.AA11834@kannix.cubiculum.com> > The other school of thought says that direct access makes code more > efficient (since you avoid a message send every time you get or set an > instance variable) and easier to read. But the main complaint seems to > be that writing accessor methods for every instance variable opens up the > object to inspection and manipulation which undermines the encapsulation > of its data. Actually, there's nothing wrong with exposing STATE of an object, which is what accessor methods do. The problem with exposing VARIABLES, i.e. breaking encapsulation, is that accessing them directly can put the object into an ill-defined state. However, if all variable access is done consequently with accessor methods, then your methods can do the proper pre- and postcondition checks, assertions, side effects, etc. In other words, proper accessor methods exactly PREVENT what is "evil" about breach of encapsulation. As such, accessor methods, if programmed properly, are very, clean and "good". But they can lead to quite some runtime overhead. Of course, for all this to work, ivars should only be used for meaningful object state, not as a computation scratchpad. That's what local variables, etc. are useful for. Needless to say, there's always a real-world conflict between academically clean programming, and performance trade-offs. As long as you are aware of the competing goal and use some reasonably good judgement, you'll probably end up with code that's fine. As others have pointed out, accessor methods have also the advantage of being extended in subclasses, etc. As for hiding accessors that should remain private: put them into a category with a private .h file. My personal approach to this is: use accessors wherever possible, then check for performance critical parts of the code, mark them with proper comments, and use direct variable access there. Just my $0.02... Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From jean-michel.cazaux at finindev.com Mon Mar 22 02:00:33 1999 From: jean-michel.cazaux at finindev.com (Jean-Michel Cazaux) Date: Thu Nov 3 14:41:54 2005 Subject: 'Washed' colors in NSImageView under NT Message-ID: <01BE7453.373A0750@JEAN-MICHEL> Hi folks, A small problem to start a new week... Under NT, bitmaps displayed in NSImageViews appears with 'washed' colors (I mean colors are not that bright and flashy they should be). I saw this because in our apps we've got a splash.bmp ; it appears really good when the app starts, but we also display it (the same file) in our custom about panel, in a NSImageView, and then colors are really 'whiter' than they were in the splash, although it's the same image. Is there something special to perform on NSImage or NSImageViews regarding the platform ? Thanks in advance for any help. __________________________________________________________________________________________ Jean-Michel CAZAUX FININDEV, Conseil en Finances pour les Collectivit?s Locales. 204 Rue Michel Teule - ZAC d'Alco 34080 Montpellier - FRANCE. T?l. +33 (0)4 67 63 66 25 - Fax +33 (0)4 67 63 35 45 e-mail: jean-michel.cazaux@finindev.com From Malcolm_Crawford at plsys.co.uk Mon Mar 22 02:35:16 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:54 2005 Subject: 'Washed' colors in NSImageView under NT In-Reply-To: <01BE7453.373A0750@JEAN-MICHEL> References: <01BE7453.373A0750@JEAN-MICHEL> Message-ID: <199903221035.KAA02561@jasmine.plsys.co.uk> Jean-Michel a ecrit: > Under NT, bitmaps displayed in NSImageViews appears with 'washed' colors (I > mean colors are not that bright and flashy they should be). I saw this > because in our apps we've got a splash.bmp ; it appears really good when > the app starts, but we also display it (the same file) in our custom about > panel, in a NSImageView, and then colors are really 'whiter' than they were > in the splash, although it's the same image. > I wonder if it's a gamma thing... cf: http://www.plsys.co.uk/~malcolm/NEXTSTEP/WWW/ and look for the section on "Gamma correction (or "Why do my images look too dark on PCs"?)". This relates to a question Wassim asked a few days ago (I can't remember which list) about gamma values on NT. I do know that it's possible to adjust the gamma values on NEXTSTEP/OPENSTEP using the currentframebuffertransfer function, and I've written a small app to allow you to change the gamma value with a slider (including for independently for RGB), however the function doesn't work under NT, which is a little confusing. Now that I have some time in the office (I was in Lugano for most of last week, so a special "Hello" here to the most generous Matt "We did the raclette thing" Kerr whom I met there) I'll try to track it down. Best wishes, mmalc. From anders at milkweed.com Mon Mar 22 04:52:25 1999 From: anders at milkweed.com (Anders Pytte) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <199903220750.AA11834@kannix.cubiculum.com> Message-ID: Ronald wrote: >> The other school of thought says that direct access makes code more >> efficient (since you avoid a message send every time you get or set an >> instance variable) and easier to read. But the main complaint seems to >> be that writing accessor methods for every instance variable opens up the >> object to inspection and manipulation which undermines the encapsulation >> of its data. > >Actually, there's nothing wrong with exposing STATE of an object, which >is what accessor methods do. The problem with exposing VARIABLES, i.e. >breaking encapsulation, is that accessing them directly can put the >object into an ill-defined state. Except if you don't want clients to make assumptions based on that state. The more of a server that is made public, the harder it is to maintain the service and change the way it works without breaking clients. >However, if all variable access is done consequently with accessor >methods, then your methods can do the proper pre- and postcondition >checks, assertions, side effects, etc. In other words, proper accessor >methods exactly PREVENT what is "evil" about breach of encapsulation. >As such, accessor methods, if programmed properly, are very, clean >and "good". But they can lead to quite some runtime overhead. >Of course, for all this to work, ivars should only be used for >meaningful object state, not as a computation scratchpad. That's >what local variables, etc. are useful for. >Needless to say, there's always a real-world conflict between >academically clean programming, and performance trade-offs. >As long as you are aware of the competing goal and use some reasonably >good judgement, you'll probably end up with code that's fine. > >As others have pointed out, accessor methods have also the advantage >of being extended in subclasses, etc. > >As for hiding accessors that should remain private: put them into a >category with a private .h file. > >My personal approach to this is: use accessors wherever possible, >then check for performance critical parts of the code, mark them >with proper comments, and use direct variable access there. There is really no use in "wherever possible". It should be always (except in constructors and destructors and the accessors themselvs) or you are just wasting your time. If you neglect using an accessor in even one case, then the accessor cannot be effectively overridden (or the deriveed class will also need to override every method of the base class that accesses the data directly, which kind of defeats the purpose). But in general I agree with Ronald. Anders. ---------------------------------------------------------------- Anders Pytte Milkweed Software Ferguson Hill Voice: (802) 472-5142 Cabot VT 05647 Internet: anders@milkweed.com ---------------------------------------------------------------- From zmonster at heliosgroup.com Mon Mar 22 08:23:41 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:54 2005 Subject: Setting up an ODBC data source on multiple NT machines Message-ID: <5948F7268092D211B2020000D11C0FB8081793@chi-serv2.heliosgroup.com> Hi, Forgive this OpenStep 4.2 NT related question, but I'm not sure what omnigroup mailing list to use anymore for OpenStep 4.2 stuff (in fact, this is probably an NT sys-admin issue, so I'll probably end up copying some NT newsgroup too). Anyway, does anyone know if there is a quick, general, one-time way to install a new ODBC data source configuration on MULTIPLE NT machines (i.e. over 100) WITHOUT having to walk to each and every machine and go through the process of setting the data source up manually on EACH machine!? I've researched this issue and I can't seem to find a way to do this - Microshit doesn't seem to provide tools for this type of "network" setup, and even if there were tools, I'm not sure they would properly update the registries on the respective machines. Keep in mind that IF there is no solution to the above problem, then I can state, without hesitation, that NT CANNOT and SHOULD NOT be used in a production environment at ANY company for the simple reason that maintaining database-driven applications on large networks is a complete impossibility! Eric From fabienr at ncmi.com Mon Mar 22 08:53:29 1999 From: fabienr at ncmi.com (Fabien Roy) Date: Thu Nov 3 14:41:54 2005 Subject: Setting up an ODBC data source on multiple NT machines In-Reply-To: <5948F7268092D211B2020000D11C0FB8081793@chi-serv2.heliosgroup.com> References: <5948F7268092D211B2020000D11C0FB8081793@chi-serv2.heliosgroup.com> Message-ID: <199903221654.IAA04333@ignem.omnigroup.com> You wrote: > Keep in mind that IF there is no solution to the above problem, > then I can state, without hesitation, that NT CANNOT and SHOULD NOT > be used in a production environment at ANY company for the simple > reason that maintaining database-driven applications on large > networks is a complete impossibility! You could use a samba server and put your ODBC stuff on a shared drive. --- Fabien L Roy NationsBanc Montgomery Securities, LLC voice: (704)386-75-76 100 North Tryon Street fax: (704)388-95-64 NC1-007-09-08 Charlotte NC 28255 Beeper: 143-9722 (1-800-946-46-46) or http://www.MobileComm.com/message/ email: fabienr@ncmi.com (NeXT/Mime) Pager-email 1439722@mobilecomm.net From zmonster at heliosgroup.com Mon Mar 22 09:11:37 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:54 2005 Subject: Setting up an ODBC data source on multiple NT machines Message-ID: <5948F7268092D211B2020000D11C0FB8081795@chi-serv2.heliosgroup.com> > -----Original Message----- > From: Fabien Roy [mailto:fabienr@ncmi.com] > Sent: Monday, March 22, 1999 11:03 AM > To: Multiple recipients of list > Subject: Re: Setting up an ODBC data source on multiple NT machines > > You could use a samba server and put your ODBC stuff on a > shared drive. I'm not sure what you mean here? "System" ODBC data sources are stored in the registry (as far as I know). Are you implying that the ODBC configurations are file-system based, and can thusly be shared over a network? Eric From fabienr at ncmi.com Mon Mar 22 09:57:19 1999 From: fabienr at ncmi.com (Fabien Roy) Date: Thu Nov 3 14:41:54 2005 Subject: Setting up an ODBC data source on multiple NT machines In-Reply-To: <5948F7268092D211B2020000D11C0FB8081795@chi-serv2.heliosgroup.com> References: <5948F7268092D211B2020000D11C0FB8081795@chi-serv2.heliosgroup.com> Message-ID: <199903221757.JAA16252@ignem.omnigroup.com> You wrote: > I'm not sure what you mean here? "System" ODBC data sources are > stored in the registry (as far as I know). Are you implying that > the ODBC configurations are file-system based, and can thusly be > shared over a network? > > Eric Sorry, I got it wrong. I was thinking about the sql.ini file. Fabien. --- Fabien L Roy NationsBanc Montgomery Securities, LLC voice: (704)386-75-76 100 North Tryon Street fax: (704)388-95-64 NC1-007-09-08 Charlotte NC 28255 Beeper: 143-9722 (1-800-946-46-46) or http://www.MobileComm.com/message/ email: fabienr@ncmi.com (NeXT/Mime) Pager-email 1439722@mobilecomm.net From rcfa at cubiculum.com Mon Mar 22 11:53:17 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: References: Message-ID: <199903221953.AA14477@kannix.cubiculum.com> > Except if you don't want clients to make assumptions based on that state. > The more of a server that is made public, the harder it is to maintain the > service and change the way it works without breaking clients. Sure, but that's why you can chose to vend only certain objects as the interface to a server, and use objects like NSProtocolChecker (or whatever the name is...) > There is really no use in "wherever possible". It should be always (except > in constructors and destructors and the accessors themselvs) or you are > just wasting your time. If you neglect using an accessor in even one case, > then the accessor cannot be effectively overridden (or the deriveed class > will also need to override every method of the base class that accesses the > data directly, which kind of defeats the purpose). Wherever possible was referring to places where performance penalties become prohibitive. Silly example: if you do a matrix inversion of a large matrix, and you access each matrix element during the calculations with accessor methods, things will just be a tad slow. But that's why violations of the rule must be well documented and preferrably should only be done for classes that are not meant to be subclassed or that are only used in circumstanced were source code will always be available. Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From anders at milkweed.com Mon Mar 22 14:42:04 1999 From: anders at milkweed.com (Anders Pytte) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <199903221953.AA14477@kannix.cubiculum.com> References: Message-ID: I wrote: >> There is really no use in "wherever possible". It should be always (except >> in constructors and destructors and the accessors themselvs) or you are >> just wasting your time. If you neglect using an accessor in even one case, >> then the accessor cannot be effectively overridden (or the deriveed class >> will also need to override every method of the base class that accesses the >> data directly, which kind of defeats the purpose). Ronald wrote: >Wherever possible was referring to places where performance penalties >become prohibitive. Silly example: if you do a matrix inversion >of a large matrix, and you access each matrix element during the >calculations with accessor methods, things will just be a tad slow. >But that's why violations of the rule must be well documented and >preferrably should only be done for classes that are not meant to >be subclassed or that are only used in circumstanced were source >code will always be available. Using a language that allows aggressive compiler optimizations of accessors may also solve this problem. But I think classes which contain intense computation should not make public any data actually used in the computations. A better solution is to make local or private copies of such data (using public accessors if they exist) for the purpose of computation. Then a clear and private boundary exists between the object's public facade and it's private rumblings. This also helps in multithreading situations, as clients may change an "in process" object's public data without locking or interfering with a computation, although perhaps rendering the eventual result obsolete, or causing it to abort or restart. Anders. ---------------------------------------------------------------- Anders Pytte Milkweed Software Ferguson Hill Voice: (802) 472-5142 Cabot VT 05647 Internet: anders@milkweed.com ---------------------------------------------------------------- From jknight at MIT.EDU Mon Mar 22 14:57:26 1999 From: jknight at MIT.EDU (James Knight) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: Message-ID: At 9:45 PM -0800 3/21/99, Fuzzy Logic wrote: >mind you, I come mainly from Java, where things are much more static than >Obj-C; I had forgotten than Obj-C can't necessarily optimize by inlining >accessors. still, I'm inclined to take a slight performance hit if it makes >code more maintainable, since I hate nothing so much as having to find all >the instances of something (say direct access) in a program, and change >them. :-) In Java, you can't optimize/inline away accessors either, for two reasons. The first is that unless they are marked final, a subclass could override them, and there is no information about what subclasses exist at compile time (java uses dynamic loading/linking for all classes). The second more serious problem is that if variables are private or protected, and the accessor is public, than the VM's security stuff would prevent access to the private variable if the compiler tried to optimize away the accessor into a direct variable reference. I don't believe Java should be any better than Obj-C in this respect, since it is just as dynamic on the inside (its just harder to access its 'dynamicism' from the language.) OTOH, the Just In Time compiler in Java CAN do that kind of optimization, since it has all the subclass information AND the ability to bypass the security mechanisms. When it compiles the java code into native code, if its smart enough, it can get rid of the function call completely. So if you're using a JIT with Java, then perhaps the overhead of the accessors will be minimal enough that you wouldn't notice it. -James -- You are in a maze of testy little Java VMs, all subtly diferent. From anders at milkweed.com Mon Mar 22 15:41:37 1999 From: anders at milkweed.com (Anders Pytte) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: Message-ID: >At 9:45 PM -0800 3/21/99, Fuzzy Logic wrote: >>mind you, I come mainly from Java, where things are much more static than >>Obj-C; I had forgotten than Obj-C can't necessarily optimize by inlining >>accessors. still, I'm inclined to take a slight performance hit if it makes >>code more maintainable, since I hate nothing so much as having to find all >>the instances of something (say direct access) in a program, and change >>them. :-) > >In Java, you can't optimize/inline away accessors either, for two reasons. >The first is that unless they are marked final, a subclass could override >them, and there is no information about what subclasses exist at compile >time (java uses dynamic loading/linking for all classes). The second more >serious problem is that if variables are private or protected, and the >accessor is public, than the VM's security stuff would prevent access to >the private variable if the compiler tried to optimize away the accessor >into a direct variable reference. I don't believe Java should be any better >than Obj-C in this respect, since it is just as dynamic on the inside (its >just harder to access its 'dynamicism' from the language.) >OTOH, the Just In Time compiler in Java CAN do that kind of optimization, >since it has all the subclass information AND the ability to bypass the >security mechanisms. When it compiles the java code into native code, if >its smart enough, it can get rid of the function call completely. So if >you're using a JIT with Java, then perhaps the overhead of the accessors >will be minimal enough that you wouldn't notice it. >-James Just to add emphasis to what James wrote, JIT compilation will probably be how most performance sensitive Java services will be deployed. In that case, all kinds of optimizations are possible. Now my question: care to explain how even an accessor marked final could be optimized away at all compile (to-byte-code) time? I seemed to have misplaced my JavaVM manual (<--weak excuse!) Seems to me that the client object knows nothing about the implementation of the server object's accessor at that time. Perhaps you mean at runtime? There hardly seems an advantage to optimizing then since the result of the optimization would not be stored so the optimization would need to be repeated each time the accessor is invoked. Or perhaps I have my head in a bog (as usual). Anders. ---------------------------------------------------------------- Anders Pytte Milkweed Software Ferguson Hill Voice: (802) 472-5142 Cabot VT 05647 Internet: anders@milkweed.com ---------------------------------------------------------------- From holger at wizards.de Mon Mar 22 15:55:59 1999 From: holger at wizards.de (Holger Hoffstaette) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: References: Message-ID: <199903222355.AAA01693@ragnarok.en.uunet.de> >Just to add emphasis to what James wrote, JIT compilation will probably be >how most performance sensitive Java services will be deployed. In that >case, all kinds of optimizations are possible. For *really* performance-sensitive Java stuff people will more likely resort to full-program analyzers & compilers to native code; this yields an even more impressive speed up. You lose dynaloading and a bit of introspection, but one could argue that only very few business apps do dynamic loading and introspection at all. Most people seem to have enough problems getting their botched source code soup to run at all already, not to mention correctly or even reliably. :-( Anyway, check out www.instantiations.com. As expected, they're a bunch of *really* cool Smalltalk guys fixing Java's problems. Too bad they can't fix Sun and Java's long term problems. >Now my question: care to explain how even an accessor marked final could be >optimized away at all compile (to-byte-code) time? I seemed to have The whole-program optimizer can actually look at each and every reference to the accessor, and if it's not overloaded, can dynamically replace each call (I don't quite dare say 'method invocation') with a direct ivar reference, which is of course faster. Without WPA dynamic recompilers like HotSpot will dynamically inline the accessors when possible, and not when not. According to Sun, that is. "Can't hardly wait." :-( Holger From anders at milkweed.com Mon Mar 22 16:18:29 1999 From: anders at milkweed.com (Anders Pytte) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: <199903222355.AAA01693@ragnarok.en.uunet.de> Message-ID: >>Just to add emphasis to what James wrote, JIT compilation will probably be >>how most performance sensitive Java services will be deployed. In that >>case, all kinds of optimizations are possible. > >For *really* performance-sensitive Java stuff people will more likely >resort to full-program analyzers & compilers to native code; this yields >an even more impressive speed up. You lose dynaloading and a bit of >introspection, but one could argue that only very few business apps do >dynamic loading and introspection at all. Most people seem to have enough >problems getting their botched source code soup to run at all already, >not to mention correctly or even reliably. :-( Ho, ho - thats good! >Anyway, check out www.instantiations.com. As expected, they're a bunch of >*really* cool Smalltalk guys fixing Java's problems. Too bad they can't >fix Sun and Java's long term problems. I find this very interesting. Thanks! >>Now my question: care to explain how even an accessor marked final could be >>optimized away at all compile (to-byte-code) time? I seemed to have > >The whole-program optimizer can actually look at each and every reference >to the accessor, and if it's not overloaded, can dynamically replace >each call (I don't quite dare say 'method invocation') with a direct ivar >reference, which is of course faster. > >Without WPA dynamic recompilers like HotSpot will dynamically inline the >accessors when possible, and not when not. According to Sun, that is. But how does "whole-program optimization" apply to compiling into byte-code packages or interpreting such packages? Either i am missing something or you misread James and my discussion. Anders. ---------------------------------------------------------------- Anders Pytte Milkweed Software Ferguson Hill Voice: (802) 472-5142 Cabot VT 05647 Internet: anders@milkweed.com ---------------------------------------------------------------- From jknight at MIT.EDU Tue Mar 23 00:27:21 1999 From: jknight at MIT.EDU (James Knight) Date: Thu Nov 3 14:41:54 2005 Subject: "Variable Free Programming" In-Reply-To: Message-ID: At 6:49 PM -0500 3/22/99, Anders Pytte wrote: >Now my question: care to explain how even an accessor marked final could be >optimized away at all compile (to-byte-code) time? I seemed to have >misplaced my JavaVM manual (<--weak excuse!) Seems to me that the client >object knows nothing about the implementation of the server object's >accessor at that time. You are correct on this, I was mis-thinking when I wrote that sentance. Also, there is one case where the compiler can (and some do) do inlining. If you use methods marked final, from within the same class, then the compiler can inline them without any problems, since it doesn't have to worry about mis-access, and it doesn't have to worry about the definition changing. -James -- You are in a maze of testy little Java VMs, all subtly diferent. From anderkh at texaco.com Tue Mar 23 12:25:21 1999 From: anderkh at texaco.com (Anderson, Ken H) Date: Thu Nov 3 14:41:54 2005 Subject: NXHosting with Mac OS X Server? Message-ID: Can someone verify whether NXHosting from an OpenStep machine to a Mac running OS X still works? It worked in DR2, and I still haven't received my OS X copy yet. Thanks, Ken Anderson From ctuel at apple.com Tue Mar 23 13:02:53 1999 From: ctuel at apple.com (Cliff Tuel) Date: Thu Nov 3 14:41:54 2005 Subject: NXHosting with Mac OS X Server? Message-ID: <199903232101.NAA21036@scv3.apple.com> NXHost capability is turned off by default, for security reasons, but you can turn it on by... 1) Login as root 2) Edit /etc/hostconfig, so that you have the line: MACHIPC=-UNRESTRICTED- 3) In the Preferences app under Expert, make the window server public. -- Cliff Tuel -- ctuel@apple.com Enterprise Technical Support / Apple Computer, Inc. From lavoie at cst.ca Tue Mar 23 13:14:12 1999 From: lavoie at cst.ca (Martin-Gilles) Date: Thu Nov 3 14:41:54 2005 Subject: NXHosting with Mac OS X Server? Message-ID: <199903232112.QAA00942@plexus.cst.ca> >NXHost capability is turned off by default, for security reasons, but you >can turn it on by... > >1) Login as root > >2) Edit /etc/hostconfig, so that you have the line: >MACHIPC=-UNRESTRICTED- > >3) In the Preferences app under Expert, make the window server public. Does that work both ways? Aka, can I see Mac OS X apps under my color slab? This did not work under DR2 (I cant remember what the color slab blabed about, but it meant there was an error). The specefic console messages were entertaining, though, when trying to use the BlueBox off the color slab. Martin-Gilles Lavoie | "No! Try not. Do, or do not. Mac OS programmer, CS&T | There is no try." lavoie@cst.ca | -- Yoda on error handling http://blackhole.cst.ca/ | Y2K statement: hahahaha! From kc at omnigroup.com Tue Mar 23 13:17:24 1999 From: kc at omnigroup.com (Ken Case) Date: Thu Nov 3 14:41:54 2005 Subject: NXHosting with Mac OS X Server? Message-ID: <199903232117.NAA09880@ignem.omnigroup.com> [Obligatory moderator's note: This sort of question belongs on MacOSX-admin, not MacOSX-dev. Please send any related messages to the admin list.] > Can someone verify whether NXHosting from an OpenStep machine to > a Mac running OS X still works? It worked in DR2, and I still > haven't received my OS X copy yet. It works if you enable Remote Mach IPC in Network Preferences. (Remote Mach IPC is turned off by default, since it's a potential security risk.) Ken From jdevlin at blacklab.la.asu.edu Tue Mar 23 13:38:09 1999 From: jdevlin at blacklab.la.asu.edu (John Devlin) Date: Thu Nov 3 14:41:54 2005 Subject: NXHosting with Mac OS X Server? In-Reply-To: References: Message-ID: <9903232138.AA02004@blacklab.la.asu.edu> > Can someone verify whether NXHosting from an OpenStep machine to a Mac > running OS X still works? > NXHost capability is turned off by default, for security reasons, but > you can turn it on by... Thanks! Can you use this to NXHost from a NextStep 3.3 machine as well? Jd. From rfox at neo.rr.com Tue Mar 23 15:48:27 1999 From: rfox at neo.rr.com (rfox) Date: Thu Nov 3 14:41:54 2005 Subject: Received Copies:What Channel? Message-ID: <19990324004901.AAA1885@roadrunner.neo.lrun.com@[24.93.174.131]> For those who have received copies already who were not part of the seeding process, did you order it through ADC (the 5 user version), retail, or education? I ordered mine through ADC on the 19th and was just hoping to hear that some had received it so quickly though there. Any insight is appreciated. I feel like a kid waiting for Christmas. Robert Fox From j-ochs at nwu.edu Tue Mar 23 20:37:09 1999 From: j-ochs at nwu.edu (Joshua Ochs) Date: Thu Nov 3 14:41:54 2005 Subject: Development on MacOS X Server In-Reply-To: <199903240132.RAA23636@mail-gw.pacbell.net> Message-ID: As a CS major who has the problem of being a Mac fanatic in a MS Visual Studio/Linux world, I'd love to be able to use MacOS X Server for my programming projects. I like being able to use the GUI for my day-to-day (even basic file maintenance), have semi-quick access to the MacOS (BlueBox isn't all that fast, but better than a full reboot), and of course, the UNIX underpinnings for the programming. However, I have a few questions and a request: 1) Has the C++ implementation been "fixed"? Under DR2 it was missing string.h (who needs a string class, right?), and also would complain about a missing library (I had posted in the past; right now I'm on unstable dialup and can't get the archives with the exact problem). 2) I've read reports that Perl is included in the box - is this true? What version? 3) Is there an SSH client for MOSXS yet? SSH is required for our main CS servers. DOes anyone have info on SCP (of which I know extremely little beyond it stands for secure copy - I just know that I need it :). 4) While I appreciate Stepwise's archives of uploaded software, could it be organized by category as well as by date? That way, if I'm looking for internet clients/servers or development tools, I know where to go quickly. After all, we now have a commercial audience. :) 5) Is there anywhere one can go to get started on porting UNIX software? It's a skill I'd like to add to my arsenal - rather than wait for others to port this or that, port it myself. However, it's a daunting task. All help is appreciated, - Joshua Ochs Northwestern University From pmb at apple.com Tue Mar 23 21:31:36 1999 From: pmb at apple.com (Peter Bierman) Date: Thu Nov 3 14:41:55 2005 Subject: Development on MacOS X Server In-Reply-To: Message-ID: At 8:56 PM -0800 3/23/99, Joshua Ochs wrote: >2) I've read reports that Perl is included in the box - is this true? What >version? version 5.005_02 >3) Is there an SSH client for MOSXS yet? SSH is required for our main CS >servers. DOes anyone have info on SCP (of which I know extremely little >beyond it stands for secure copy - I just know that I need it :). Various people have ported it. I'm sure someone here will reply with a URL. scp is just one of the tools installed with ssh. It's just like cp, but takes user@host:path for the arguments, and does all of its work encrypted. >5) Is there anywhere one can go to get started on porting UNIX software? >It's a skill I'd like to add to my arsenal - rather than wait for others to >port this or that, port it myself. However, it's a daunting task. It's mostly an art of luck. Find a small UNIX tool that you want. Unpack it, and run make. Look at the compiler output, and try to fix the output as if you were fixing a bug in your own program. If you're lucky, the thing you're trying to port only has 3 or 4 problems. Usually it's things like including the wrong headers. -pmb -- "UNIX shells in Mac OS X should be unneeded but functional... and have the same installed base as MPW." From sanguish at digifix.com Tue Mar 23 21:38:51 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:55 2005 Subject: Development on MacOS X Server In-Reply-To: References: Message-ID: <199903240538.AAA00730@digifix.com> Joshua Ochs wrote: > 2) I've read reports that Perl is included in the box - is this true? What > version? > Yes.. 5.005_2 > 3) Is there an SSH client for MOSXS yet? SSH is required for our main CS > servers. DOes anyone have info on SCP (of which I know extremely little > beyond it stands for secure copy - I just know that I need it :). > scp comes with ssh.. I think that it is compilable though... > 4) While I appreciate Stepwise's archives of uploaded software, could it be > organized by category as well as by date? That way, if I'm looking for > internet clients/servers or development tools, I know where to go quickly. > After all, we now have a commercial audience. :) Yes.. its coming... :-) Hopefully I'll get time this weekend to work on it.. > > 5) Is there anywhere one can go to get started on porting UNIX software? > It's a skill I'd like to add to my arsenal - rather than wait for others to > port this or that, port it myself. However, it's a daunting task. > macosx-port-request@stepwise.com is a mailing list.. From dyoung at vviuh221.vvi.com Wed Mar 24 04:21:07 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:55 2005 Subject: Development on MacOS X Server References: Message-ID: <9903241221.AA06226@vviuh221.vvi.com> >>5) Is there anywhere one can go to get started on porting UNIX software? >>It's a skill I'd like to add to my arsenal - rather than wait for others to >>port this or that, port it myself. However, it's a daunting task. >It's mostly an art of luck. Find a small UNIX tool that you want. Unpack >it, and run make. Look at the compiler output, and try to fix the output as >if you were fixing a bug in your own program. If you're lucky, the thing >you're trying to port only has 3 or 4 problems. Usually it's things like >including the wrong headers. That is so true! I like to go to www.freebsd.com now and then to find a goodie for Mac OS X Server and learn something new. Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From bbum at codefab.com Wed Mar 24 06:50:06 1999 From: bbum at codefab.com (Bill Bumgarner) Date: Thu Nov 3 14:41:55 2005 Subject: Development on MacOS X Server References: <9903241221.AA06226@vviuh221.vvi.com> Message-ID: <36F8FB9E.DEA2ECA4@mail.codefab.com> The absolutely positively best resource for finding the latest revisions of unix related tools is: www.freshmeat.net Be prepared to be overwhelmed-- there are thousands of programs listed there. I also keep an eye on www.slashdot.org As major new projects are typically announced there. b.bum David Young wrote: > >>5) Is there anywhere one can go to get started on porting UNIX software? > >>It's a skill I'd like to add to my arsenal - rather than wait for others to > >>port this or that, port it myself. However, it's a daunting task. > > >It's mostly an art of luck. Find a small UNIX tool that you want. Unpack > >it, and run make. Look at the compiler output, and try to fix the output as > >if you were fixing a bug in your own program. If you're lucky, the thing > >you're trying to port only has 3 or 4 problems. Usually it's things like > >including the wrong headers. > > That is so true! I like to go to www.freebsd.com now and then to find a > goodie for Mac OS X Server and learn something new. > > Thanks A Bunch! David Young; VVI-DCS > http://www.vvi.com From Philippe.Robert at uptime.ch Wed Mar 24 06:58:29 1999 From: Philippe.Robert at uptime.ch (Philippe C.D. Robert) Date: Thu Nov 3 14:41:55 2005 Subject: *urgent*... Message-ID: <9903241458.AA00718@uptime.ch> Hi! I have some weird problems here on OS4.2 and EOF2.1 with OpenBase 5.1 If I set up an EO newReport like that: [descriptionField selectAll:self]; if([[descriptionField textStorage] containsAttachments]) { reportDescription = [descriptionField RTFDFromRange:[descriptionField selectedRange]]; } else { reportDescription = [descriptionField RTFFromRange:[descriptionField selectedRange]]; } [newReport setReportDescription:reportDescription]; // Get the initial list [editingContext insertObject:newReport]; [editingContext saveChanges]; I get this: Mar 24 15:47:05 DBurst[1037] adaptorValueByConvertingAttributeValue: -- EOAttribute 0x150ad4: unable to convert value of class NSInlineCString for attribute reportDescription in entity Report to adaptor type EOAdaptorBytesType But reportDescription definitly is a NSData...?! *Any* hints are helpful, since it must be fixed until tomorrow morning...;-) Are there any special things I have to look at in OpenBase?! In Oracle it seems to work... sweet dreams, Phil --- Philippe C.D. Robert Uptime Object Factory Inc http://www.nice.ch/~phip Software Development http://www.nice.ch/ProjectCenter OpenStep/Unix From sanguish at digifix.com Wed Mar 24 14:24:00 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:55 2005 Subject: Mac OS X Server Port list - small address correction In-Reply-To: <199903242232.RAA17120@project.lotek.org> References: <199903242232.RAA17120@project.lotek.org> Message-ID: <199903242224.RAA04595@digifix.com> Doh! That should have been macosx-port-request@digifix.com with subscribe in the subject.. From jordandm at dea-mattson.COM Wed Mar 24 17:30:33 1999 From: jordandm at dea-mattson.COM (Jordan Dea-Mattson) Date: Thu Nov 3 14:41:55 2005 Subject: Unix development resources for Mac OS X Server Message-ID: <1289797042-109620862@GraceComm-CovChurch.ORG> All - If you are looking for resources on Unix development that discusses Mac OS X Server, you might want to check out: www.macunix.com It is a great site established by Clark Shishido. Disclaimer: I went to University with Clark's older sister Bonnie and we are best of friends. Clark and I also helped to beta test the ill-fated eWorld a number of years ago. But neither of those influenced my opinion of the site. Yours, Jrodan From classical at metawire.com Wed Mar 24 08:48:46 1999 From: classical at metawire.com (Matt Covey, Classical Software) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? Message-ID: At the risk of admitting to being blind, just where on the ADC site do you actually order Mac OS X Server? After logging in, all I see are two pages of tech support, etc ("Purchases"). The only mention I can find is in the popup Help window. (And yes, I'm have an ADC Select membership - my profile comfirms it. [Before this year I was an Associate Developer for more years than I care to remember.]) Thanks for any insight, ++matt ("puzzled and wanting to order") covey From td at twics.com Wed Mar 24 18:13:58 1999 From: td at twics.com (Troy Dawson) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? In-Reply-To: Message-ID: On Wed, 24 Mar 1999, Matt Covey, Classical Software wrote: > At the risk of admitting to being blind, just where on the ADC site do you actually order Mac OS X Server? >From the Purchase screens of your account information page. Screen 1 is ADC levels, Screen 2 is other products, like 5-client and the QT SDK. =td= From aswift at redmud.com Wed Mar 24 17:25:52 1999 From: aswift at redmud.com (Adam Swift) Date: Thu Nov 3 14:41:55 2005 Subject: FIXED> Only VGA-resolution on Blue-White G3 Message-ID: <36F9909F.377A726@redmud.com> I just got Mac OS X Server and performed the install onto a vanilla Blue & White G3. Installation was a snap, but I was stuck with 640x480 VGA resolution on a system which WAS running 1600x1200 under 8.5.1 on my ViewSonic VS790 MacOS compatible monitor. When I called Apple for help, I was told there was no phone support for OS X Server, however the support guide that shiped with it stated that you have 90 days of phone support for "installation support". I'm not sure if I misunderstood what is covered by that or if the help desk has not been informed. They recommended looking at the site: http://www.info.apple.com/support/macosxserver/ Which did in fact lead me to the solution (posted about 3 hours AFTER I was told to look there!) The solution is trivial (as posted by Alexander Hinds on http://discuss.info.apple.com/boards/osxservr.nsf/by+topic?OpenView) Enter: defaults write Preferences NSEnableShowAllDisplayModes YES Now if you run Preferences.app, you will be able to choose to view "All" possible resolutions, instead of getting stuck with the only recommended option: 640x480. Thank you VERY much Mr. Hinds! - adam P.S. Apologies for cross posting to macosx-dev, but it's REALLY hard to develop in 640x480 mode. From khan at mediaaccess.com Thu Mar 25 01:08:11 1999 From: khan at mediaaccess.com (Khan Klatt) Date: Thu Nov 3 14:41:55 2005 Subject: Driver for Adaptec 2930 Card? Message-ID: Does anyone know how to get a 2930 card working in MacOS X Server? I've also a 2906 card handy, but since it doesn't have any BIOS, I expect it's a little harder to support. I saw on Apple's site that one user mentioned that the 2940 was not usable, but that the 2930, somehow, was. Does anyone have any idea how I can use my SCSI drive with MacOS X Server? Details of the apple site user post below. URL: Post: >I received a response from Adaptec Support a few moments ago: > >"Unfortunately, the PowerDomain 2940UW board does not function under OS X. >The only retail >card that will work for now is the 2930U Mac card." > >I wrote back, asking if Adaptec has plans to release Mac OS X drivers >and/or an updated firmware >for the 2940UW some time in the future. I will post any response here. > > Regards, > Michael A. Lowry > mlowry@austin.rr.com -Khan From andreas at mevis.de Thu Mar 25 05:30:19 1999 From: andreas at mevis.de (Andreas Bohne) Date: Thu Nov 3 14:41:55 2005 Subject: NSBox notification? Message-ID: <9903251430.AA190231@matisse> Hi, I have subclassed NSBox and want to handle the addition of new views to the content view of an object of that class in IB. NSDraggingDestination protocol seems not to work in design mode (at least I couldn't find out how to get it to work). So I'm looking for the right point to add my code after adding a subview - something like a NSBoxDidAddSubviewToContentViewNotification. I would appreciate any suggestion. Sincerly, Andreas From janos.lobb at yale.edu Thu Mar 25 07:16:04 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? In-Reply-To: Message-ID: At 21:25 -0500 3/24/1999, Troy Dawson wrote: >On Wed, 24 Mar 1999, Matt Covey, Classical Software wrote: > >> At the risk of admitting to being blind, just where on the ADC site do >>you actually order Mac OS X Server? > >>From the Purchase screens of your account information page. > >Screen 1 is ADC levels, Screen 2 is other products, >like 5-client and the QT SDK. > Not for everyone. I have two Select accounts. Using one - xxx02176 - I was able to order MacOSX server yesterday. Using the second one -xxx62064 - today I did not see MacOSX Server on the purchasable items list. >=td= J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From potts at umich.edu Thu Mar 25 07:41:15 1999 From: potts at umich.edu (Paul R. Potts) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? References: Message-ID: <99032510422005.02875@LANmodem> I was trying to order it on the 19th and 20th and it did not show up on my account... but then magically, a day or two ago, it apeared as a purchasable item. I'm not clear why, but we did place our order! Paul On Thu, 25 Mar 1999, János Löbb wrote: > Not for everyone. I have two Select accounts. Using one - xxx02176 - I > was able to order MacOSX server yesterday. Using the second one -xxx62064 > - today I did not see MacOSX Server on the purchasable items list. > > >=td= > > > János Löbb Tel: 203-737-5204 Yale University Pathology > Fax: 203-785-7303 310 Cedar St. Room BML104A > janos.lobb@yale.edu New Haven CT 06510 > > Never take a candid-cookie from a stranger. -- -- "Sugar contains no fat." - seen on a box of sugar Paul R. Potts - potts@umich.edu - Health Media Research Lab From dyoung at vviuh221.vvi.com Thu Mar 25 08:43:27 1999 From: dyoung at vviuh221.vvi.com (David Young) Date: Thu Nov 3 14:41:55 2005 Subject: [other ordering] Re: Ordering: Just where on ADC? References: <99032510422005.02875@LANmodem> Message-ID: <9903251643.AA07711@vviuh221.vvi.com> If you need commercial copies I suggest: http://www.softwarestreet.com/bin/catalog/getProdPage.cgi?sku=121281 >From the time to order to the time of delivery it was less than 24 hours! Thanks A Bunch! David Young; VVI-DCS http://www.vvi.com From xavierb at proxyware.fr Thu Mar 25 09:14:46 1999 From: xavierb at proxyware.fr (Xavier Barrier) Date: Thu Nov 3 14:41:55 2005 Subject: MOSX Server and RAID5 Message-ID: <9903251714.AA06107@proxyware.fr> A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 973 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990325/6bbb3bd8/attachment.bin From abridge at dcn.davis.ca.us Thu Mar 25 09:15:28 1999 From: abridge at dcn.davis.ca.us (Adam Bridge) Date: Thu Nov 3 14:41:55 2005 Subject: Are paper documents available? Message-ID: <199903251715.JAA16794@wheel.dcn.davis.ca.us> Are real, hard-copy, bound documents available for WebObjects, Yellow Box, and EOF? I can't find any reference to them. If they aren't it's a tragic oversite. I know I can print the documentation -- but I can only print single-sided and I can't bind it. So I get documentation that I simply can't use well. YES, I know it's available on-line. But I'm talking about using it when I either don't have my computer handy or I'm riding in an airplane when there's not enough room to read the damn stuff. (don't get me started about cattle-cars....pant pant pant) Thanks Adam Bridge From winstonNOSPAM at HEBB.scilearn.com Thu Mar 25 10:39:04 1999 From: winstonNOSPAM at HEBB.scilearn.com (Winston Wolff) Date: Thu Nov 3 14:41:55 2005 Subject: MOSX Server and RAID5 References: <9903251714.AA06107@proxyware.fr> Message-ID: <36FA82C8.4FA7D99C@scilearn.com> Xavier- RAID5 can be implemented in software or hardware so you are half right. I think mostlly PCs have hardware Raid systems. The Sun and Apple boxes I've seen with raids are all software. -winston ------------- Xavier Barrier wrote: > Hi there! > I'm confused about the lack of RAID5 support for MOSX. > To my understanding, RAID5 is not an OS issue but an hardware one. > I mean to say, Apple has nothing to do to support RAID5 far from supporting a RAID5 SCSI Card like some sold by Adaptec. > So, either i miss the point and RAID5 *is* something to be handled by MOSX or I miss Apple's Marketing. > Because i can't understand a position that is not to integrate a Server needed feature (RAID5) into a server aimed OS. > If any body got the answer i would feel less stupid! > Oh BTW any one knows when EAP members will be shipped with MOSX (in France!) > PS : in the case this post should not be done into this list, please forgive me! --- > Xavier BARRIER --------------------------------------------------------- ProxyWare SARL 8, rue de l'Isly 75008 Paris France Financial Software Company > Tel : (+33) 1 44 69 83 06 Fax : (+33) 1 44 69 06 92 e-mail : xavierb@proxyware.fr From cdoherty at skidmore.edu Thu Mar 25 11:27:19 1999 From: cdoherty at skidmore.edu (Fuzzy Logic) Date: Thu Nov 3 14:41:55 2005 Subject: NeXTMail format? Message-ID: hi-- premise: writing an application that handles NeXTMail. is NeXTMail just RTF tarred, compressed, and uuencoded? if not, is there an actual specification for it? and, the next question: is there a facility, either in YB or that someone else has written, for handling NeXTMail? or would I have to either roll my own or figure out how to use the Unix utilities? any help appreciated. I've checked all the mailing list archives without success. thanks, chris ------------------------------- Chris Doherty cdoherty@skidmore.edu If you cannot find the truth right where you are, where else do you expect to find it? -- Dogen It is from moments of confusion that we develop moments of clarity. From agusn at purdue.edu Thu Mar 25 13:18:58 1999 From: agusn at purdue.edu (Agus Nugroho H) Date: Thu Nov 3 14:41:55 2005 Subject: [Newbie] How to use Japanese Input Method? Message-ID: <36FAA842.75A8FC97@purdue.edu> Greetings, all. I just bought WO 4.0 for NT, and after playing for a while, I found a directory for InputManagers which include "MS-Windows Japanese Input". I have some simple questions that I cannot find the answers in InfoCenter: 1. How can I use this input method in an application? 2. Can I use it just like Microsoft Global IME for IE? 3. Can I localize an application in WO 4/NT? (the Language field in Project Inspector is uneditable) thanks in advance, agusn From Malcolm_Crawford at plsys.co.uk Thu Mar 25 13:28:59 1999 From: Malcolm_Crawford at plsys.co.uk (mmalcolm crawford) Date: Thu Nov 3 14:41:55 2005 Subject: NeXTMail format? In-Reply-To: References: Message-ID: <199903252129.VAA02210@jasmine.plsys.co.uk> Chris wrote: > is NeXTMail just RTF tarred, compressed, and uuencoded? if not, is there > an actual specification for it? > Tarred, compressed, and uuencoded RTF, or RTFD for mail with "attachments". However to the best of my knowledge there was never an official specification released. > and, the next question: is there a facility, either in YB or that someone > else has written, for handling NeXTMail? > Umm, in what sense "handle" NeXTMail? MailViewer.app reads NeXTMail. It only sends ASCII and MIME, but you don't lose your "legacy" mailboxes. Many thanks to Apple for this. > or would I have to either roll my > own or figure out how to use the Unix utilities? > If you want to convert existing NeXTMail to MIME that's a different issue -- I'm sure that I've heard of someone, somewhere having written a script to do this, and would love to be reminded of who and where! Best wishes, mmalc. From malte at oops.se Thu Mar 25 13:56:28 1999 From: malte at oops.se (Malte Tancred) Date: Thu Nov 3 14:41:55 2005 Subject: NeXTMail format? In-Reply-To: <199903252129.VAA02210@jasmine.plsys.co.uk> Message-ID: Hi, mailapputilities can be downloaded from ftp.next.peak.org I believe. Nice utils for managing NeXTmail mailboxes. Don't know about it's composing and converting abilities though. Cheerio, Malte -- Malte Tancred, OOPS ab malte@oops.se http://www.oops.se On Thu, 25 Mar 1999, mmalcolm crawford wrote: ... > If you want to convert existing NeXTMail to MIME that's a different issue -- > I'm sure that I've heard of someone, somewhere having written a script to do > this, and would love to be reminded of who and where! ... From herren at flannet.middlebury.edu Thu Mar 25 13:00:47 1999 From: herren at flannet.middlebury.edu (David Herren) Date: Thu Nov 3 14:41:55 2005 Subject: [Newbie] How to use Japanese Input Method? References: <36FAA842.75A8FC97@purdue.edu> Message-ID: agusn@purdue.edu writes: >I just bought WO 4.0 for NT, and after playing for a while, I found a >directory for InputManagers which include "MS-Windows Japanese Input". >I have some simple questions that I cannot find the answers in >InfoCenter: >1. How can I use this input method in an application? >2. Can I use it just like Microsoft Global IME for IE? >3. Can I localize an application in WO 4/NT? > (the Language field in Project Inspector is uneditable) I've been told that input methods are broken under WOF4. We haven't been able to get them to work at all. IF you discover anything, I'd LOVE to hear about it and I'll shoot any information I get your way. -- David D. Herren www.cet.middlebury.edu/herren Assoc. Dir. for Tech. & Instruction herren@flannet.middlebury.edu Center for Educational Technology voice: (802)443-5746 5 Court Street, Middlebury, VT 05753 fax: (802)443-2053 From kc at omnigroup.com Thu Mar 25 13:58:00 1999 From: kc at omnigroup.com (Ken Case) Date: Thu Nov 3 14:41:55 2005 Subject: NeXTMail format? Message-ID: <199903252158.NAA14888@ignem.omnigroup.com> > is NeXTMail just RTF tarred, compressed, and uuencoded? That's the basic idea. However, the RTF has some keywords that are non-standard, even for NeXT Text objects. > if not, is there an actual specification for it? No, unfortunately. > and, the next question: is there a facility, either in YB or that > someone else has written, for handling NeXTMail? I've written some Foundation- and AppKit-based code which does this (using the 3.3 Foundation and AppKit). It would take a little bit of work to get this going under OPENSTEP, but probably not too much. (It should be a lot easier now, since they've made the wrapper stuff in Foundation public. In 3.3, I had to hack around various private classes.) If you're interested, I'll try to dig it up and release it under Omni's standard source license. Ken From cmh at greendragon.com Thu Mar 25 17:37:38 1999 From: cmh at greendragon.com (Chris Hanson) Date: Thu Nov 3 14:41:55 2005 Subject: Does Mac OS X Server support driver development? Message-ID: I know IOKit is going to replace DriverKit someday soon. Even so, does Mac OS X Server include the necessary header files, libraries, documentation, and (dare I ask) examples for developing DriverKit device drivers? From pmb at apple.com Thu Mar 25 17:58:24 1999 From: pmb at apple.com (Peter Bierman) Date: Thu Nov 3 14:41:55 2005 Subject: Does Mac OS X Server support driver development? In-Reply-To: Message-ID: >Even so, does Mac OS X Server include the necessary header files, >libraries, documentation, and (dare I ask) examples for developing >DriverKit device drivers? http://publicsource.apple.com/ How much more could you want? -pmb -- "UNIX shells in Mac OS X should be unneeded but functional... and have the same installed base as MPW." From wsanchez at apple.com Thu Mar 25 20:06:32 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:55 2005 Subject: Does Mac OS X Server support driver development? Message-ID: <199903260406.UAA18868@scv1.apple.com> | Even so, does Mac OS X Server include the necessary header files, | libraries, documentation, and (dare I ask) examples for developing | DriverKit device drivers? DriverKit source isn't up on the web site yet, but it will be, and that will be sufficient to write and build drivers. As for examples, see the drivers in the kernel for PPC, and any others we start posting up. The documentation is the source code. We can't ask the driver team to provide anything more than that, since they need to be working on IOKit. -Fred From ttakeo at amtec.co.jp Thu Mar 25 20:47:25 1999 From: ttakeo at amtec.co.jp (tetuya takeo) Date: Thu Nov 3 14:41:55 2005 Subject: [Newbie] How to use Japanese Input Method? In-Reply-To: <36FAA842.75A8FC97@purdue.edu> References: <36FAA842.75A8FC97@purdue.edu> Message-ID: <9903260439.AA07547@amtec000.amtec.co.jp> Hi, > I just bought WO 4.0 for NT, and after playing for a while, I found a > directory for InputManagers which include "MS-Windows Japanese Input". > > I have some simple questions that I cannot find the answers in > InfoCenter: > 1. How can I use this input method in an application? I use WO 4.0 with NT 4.0 Japanese version, so I don't know "MS-Windows Japanese Input" works well with other language versions. At least, on Japanese version, it works. If you install WO 4.0 on NT Japanese version and make IME enable, choose Custom Install option in installer app, and check IME option in localization. > 3. Can I localize an application in WO 4/NT? > (the Language field in Project Inspector is uneditable) > I edited PB.project file's LANGUAGE property directly by an editor. It seems to have no problem. I don't know why Project Inspector inhibits to change language field, whether is it bug or intentional one. Cheers, --- Tetuya TAKEO ttakeo@amtec.co.jp (NeXTmail/MIME accepted) From bungi at omnigroup.com Thu Mar 25 21:09:12 1999 From: bungi at omnigroup.com (Timothy J. Wood) Date: Thu Nov 3 14:41:55 2005 Subject: Does Mac OS X Server support driver development? Message-ID: <199903260509.VAA00872@mungi.omnigroup.com> >I know IOKit is going to replace DriverKit someday soon. > >Even so, does Mac OS X Server include the necessary header files, >libraries, documentation, and (dare I ask) examples for developing >DriverKit device drivers? It includes the headers you need, but no documentation. DriverKit hasn't changed substantially from NEXTSTEP 3.3, so if you have access to that, you can use the documentation from there. Apple hasn't supplied the docs to DriverKit in the release since it is deprecated, but it does work (I thought there was something on their web site, but I can't find it right at the moment). I have written several drivers, including a driver for the Voodoo2, under MacOS X Server, so once you get source to DriverKit, it should be possible to write drivers yourself. -tim From raoul at mobopro.com Thu Mar 25 21:41:00 1999 From: raoul at mobopro.com (Raoul Duke) Date: Thu Nov 3 14:41:55 2005 Subject: NEXTSTEP 3.3 -> Mac OS X Server Message-ID: <36FB1DEC.79FBA5E6@mobopro.com> I have a couple of NEXTSTEP programs that would be useful to me on Mac OS X Server. It would take quite a bit of work to get them on Mac OS X Server, since they aren't even OpenStep-compliant. By chance, are those old conversion utilities--that were part of OPENSTEP for Mach, I believe--part of Mac OS X Server? Since they most likely aren't, would Apple consider releasing them to the NEXTSTEP community, unsupported? (There might be other old throwbacks like me.) Whom should I contact? From rthomas at uiuc.edu Thu Mar 25 21:56:48 1999 From: rthomas at uiuc.edu (Ryan Thomas) Date: Thu Nov 3 14:41:55 2005 Subject: ?: Mixing Mac OS X Server & NT Server Message-ID: Sorry for the cross-posting, but my search through the archives indicated that it's been a topic among macosx-dev subscribers ("login hooks vs custom authenticators" back in mid-'98) -- but I'm not a developer. I'm looking for a product or an API to reduce the redundancy of user account management on disparate NOSes. I've got a programmer on hand if necessary: What are the available products to allow NT & UNIX servers to use a common authentication service (Kerberos, ADS, NDS, NIS, NetInfo, etc.)? My understanding is that UNIX can do kerberos, NIS, & NetInfo (OSX)...but NT only does NT & NDS (& NIS, but requires corresponding account in local SAM of every NT box accessed by user). So, is there something else? Another way? Perhaps some way of synchronizing users/groups like Microsoft did with NT & Netware? What about Mac OS X being able to use NDS or NT/ADS? Should I just use kerberos & hope NT5 will fit in as advertised? Thanks for any help. -Ryan PS I'm aware of the code available in MacOSX for kerberos -- but it's my understanding NT doesn't support it fully...yet. NT5, right? I'm aware of SAMBA for exporting UNIX directories via SMB -- but that only covers SMB clients. I'm also aware of NISGINA for NT allowing an NT box to use NIS for authentication, but my understanding is each NT machine still needs to have a matching account in its local SAM to support a corresponding NIS login. ===========------------------------------------------------------- Ryan Thomas, CCSO Consultant Network & Systems Administrator for the College of Education, UIUC rthomas@uiuc.edu, 217/244-0429, http://www.uiuc.edu/ph/www/rthomas From penrose at sfc.keio.ac.jp Thu Mar 25 22:27:36 1999 From: penrose at sfc.keio.ac.jp (Christopher Penrose) Date: Thu Nov 3 14:41:55 2005 Subject: NEXTSTEP 3.3 -> Mac OS X Server References: <36FB1DEC.79FBA5E6@mobopro.com> Message-ID: <36FB28CA.967DE137@sfc.keio.ac.jp> Raoul Duke wrote: > > I have a couple of NEXTSTEP programs that would be useful to me on Mac OS X > Server. It would take quite a bit of work to get them on Mac OS X Server, since > they aren't even OpenStep-compliant. > > By chance, are those old conversion utilities--that were part of OPENSTEP for > Mach, I believe--part of Mac OS X Server? Since they most likely aren't, would > Apple consider releasing them to the NEXTSTEP community, unsupported? (There > might be other old throwbacks like me.) Whom should I contact? Though there may be others on this list who disagree with me, the conversion programs were not useful for me. I have ported two complex applications from NeXTstep 3.x to Openstep 4.x, and I strongly recommend doing the port strictly in a manual fashion. I was working on one of them last night even, because the memory management is still not quite right. But since I am running Openstep 4.1, it is difficult to know if these are my bugs or NeXT's (still not Apple's) NSZone problems. The conversion routines are not smart enough to make all changes automatically without introducing flaws because of the subtleties of object usage. Many of the changes in the API are simple, like the change from (id) to (void) as the standard API return type, but they don't necessarily have simple ramifications in your code, particularly if your code relies on receiving an object from the API that no longer returns one. For this reason alone (and I haven't even begun to talk about differences between NSObject and Object based classes) it is a good thing to work through the changes by hand. NXView (now NSView) has some significant changes too, that the conversion programs will not help with if you have subclassed NXView's drawSelf: method. It is now drawRect:, which requires some changes in implementation (but the resulting code should actually be simpler). The port to the Openstep API from NeXTstep 3.x is not trivial, particularly if you are not using the Foundation Kit (it was available in 3.3), but it is much easier than a rewrite. NeXTstep 3.x to Openstep 4.x is a far bigger leap than Openstep 4.x to MacOS X Server. Christopher penrose@sfc.keio.ac.jp From eric at alum.mit.edu Thu Mar 25 23:04:37 1999 From: eric at alum.mit.edu (Eric Hermanson) Date: Thu Nov 3 14:41:55 2005 Subject: NEXTSTEP 3.3 -> Mac OS X Server References: <36FB28CA.967DE137@sfc.keio.ac.jp> Message-ID: <36FB3185.3121FB29@alum.mit.edu> Christopher Penrose wrote: > Though there may be others on this list who disagree with me, the > conversion programs were not useful for me. I disagree. :) > I have ported two complex > applications from NeXTstep 3.x to Openstep 4.x, and I strongly recommend > doing the port strictly in a manual fashion. > The conversion routines are not smart enough to make all changes > automatically without introducing flaws because of the subtleties of > object usage. I've also ported several complex applications from NEXTSTEP to OpenStep 4.2 (actually, I've ported 16!). I've used both methods - by hand and the conversion scripts. With no exceptions, I was able to fully port (and test) the applications which were ported by conversion script 3-5 times faster than doing it manually. Yes, the previous author is correct in the assumption that using the conversion scripts will introduce some unexpected "bugs". However, the porting process will always include an in-depth testing/bug-fix phase (no matter what method you choose), and all of these issues can easily be ironed out during testing. For porting apps, here are some recommendations: - Read, re-read, and then read again ALL release notes for OpenStep 4.2 before you do anything. There is a ton of valuable information on tips for porting. Also spend at least a week reading all FoundationKit, AppKit, and EOF2.x documentation to get a real feel for how OpenStep works - it's an extremely complex (yet elegant) system. The more you read before the port, the quicker you'll get the port done, it's that simple. - Use the conversion scripts to port your code from NEXTSTEP to OpenStep FIRST (i.e. phase 1) - save the EOF 1.x --> 2.x port for last. Get your OpenStep/EOF1.x application working identically to the NEXTSTEP/EOF1.x app before doing the EOF port. Be extra careful about NSTableView related classes - the table view changed dramatically, as did the way several delegate methods work on classes like NSTableView, NSTextField, etc. - During the EOF phase of your port, make sure you fully understand the differences between the "Data Source" paradigm of EOF1.x (i.e. having to manually manage inserts, updates, etc), and the "EOEditingContext" paradigm of EOF2.x (i.e. updates broadcast to the editing context automatically when accessor methods are called, and ALL changes are saved when saveChanges are called). Other things to be careful about are EOF2.x's support for nested editing context's (and the ability to have fully independent documents with their own object-spaces) - many database-operation related things will behave much differently in EOF2.x than they did in 1.x. EOF2.x also has support for auto-primary-key generation and other nice data-manipulation if you're EOModel is set up correctly - see the release notes. - Eric From rcfa at cubiculum.com Thu Mar 25 22:50:50 1999 From: rcfa at cubiculum.com (Ronald C.F. Antony) Date: Thu Nov 3 14:41:55 2005 Subject: Does Mac OS X Server support driver development? In-Reply-To: <199903260406.UAA18868@scv1.apple.com> References: <199903260406.UAA18868@scv1.apple.com> Message-ID: <199903260650.AA18228@kannix.cubiculum.com> > DriverKit source isn't up on the web site yet, but it will be, and > that will be sufficient to write and build drivers. What are the chances that the drivers for black hardware will be released on an AS-IS basis? (I know they are not DriverKit ready, but they would contain enough information to write/port DriverKit to black hardware...) Also, being able to reuse my black hardware for some network gruntwork tasks would be rather useful given that we have to get all new machines for production work anyhow. Ronald ============================================================================== "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." G.B. Shaw | rcfa@cubiculum.com | NeXT-mail welcome From sanguish at digifix.com Fri Mar 26 00:59:32 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:55 2005 Subject: NEXTSTEP 3.3 -> Mac OS X Server In-Reply-To: <36FB1DEC.79FBA5E6@mobopro.com> References: <36FB1DEC.79FBA5E6@mobopro.com> Message-ID: <199903260859.DAA16832@digifix.com> Raoul Duke wrote: > I have a couple of NEXTSTEP programs that would be useful to me on Mac OS X > Server. It would take quite a bit of work to get them on Mac OS X Server, > since they aren't even OpenStep-compliant. > > By chance, are those old conversion utilities--that were part of OPENSTEP > for Mach, I believe--part of Mac OS X Server? Nope. > Since they most likely > aren't, would Apple consider releasing them to the NEXTSTEP community, > unsupported? (There might be other old throwbacks like me.) Whom should I > contact? You're probably better off trying to find someone with OS 4.2 willing to run them for you.. From andreas at mevis.de Fri Mar 26 03:49:48 1999 From: andreas at mevis.de (Andreas Bohne) Date: Thu Nov 3 14:41:55 2005 Subject: NSBox subclass on IBPalette - adding subviews in IB Message-ID: <9903261249.AA481772@matisse> We want to make a subclass of NSBox available in IB on a palette. In a previous posting to this list I asked how to know when a subview is added to my NSBox subclass object in IB. The problem is that IB does not send addSubview to the NSBox subclass object but instead directly to its contentView (encapsulation?) - so the NSBox subclass does not have a chance to handle this. It turned out that this is a known problem according to Ken's kind answer. Because I couldn't find a good solution, I will describe the problem a little more clearly hoping that it sounds familiary to someone. I want to have an NSBox like object that can get some subviews added at design time in IB, with some additional functionality (like arranging the views in specific manner). So I subclassed NSBox and put that subclass on a palette in IB. Now I want to have executed some of my code each time a subview is added. Because subviews are added not to NSBox itself but to its content view (and NSBox gets no notification about an added subview), I overwrote addSubview: in an NSView derived class and added my code here. In initialization, I replaced NSMyBox' contentView with that view. Unfortunately, in Ken's words, "IB isn't happy when you change that content view" what results in not being able to focus it in IB by double-clicking. This is somewhat not the expected behaviour, so may be someone else has seen that and found a workaround. Thanks a lot, Andreas -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 1755 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990326/52e7d57b/attachment.bin From nberch at db.lv Fri Mar 26 06:16:08 1999 From: nberch at db.lv (Nils Berzins) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? (Off topic !) Message-ID: <199903261418.QAA26909@saulite> >I was trying to order it on the 19th and 20th and it did not show up on my >account... but then magically, a day or two ago, it apeared as a >purchasable item. I'm not clear why, but we did place our order! > >Paul Hi, It seems that there are some problems with "conect.apple.com". One week ago I tried to log in to my account and register for wwdc99 but was presented with a page that our account has been disabled. I sent already two emails to devprograms@apple.com but so far got no response. I just hope that they sort this out before 15th of April... Best Regards Nils Berzins IT Manager Diena-Bonnier Ltd. From markm at mail.tyrell.com Fri Mar 26 06:53:19 1999 From: markm at mail.tyrell.com (Mark F. Murphy) Date: Thu Nov 3 14:41:55 2005 Subject: Ordering: Just where on ADC? (Off topic !) In-Reply-To: <199903261418.QAA26909@saulite> Message-ID: At 6:21 AM -0800 3/26/99, Nils Berzins wrote: >It seems that there are some problems with "conect.apple.com". One week >ago I tried to log in to my account and register for wwdc99 but was >presented with a page that our account has been disabled. I sent already >two emails to devprograms@apple.com but so far got no response. I just >hope that they sort this out before 15th of April... I called Apple Developer Relations (800-633-2152) to get mine taken care of. mark --------------------------------------------------------------------------- Mark F. Murphy, Director Software Development Tyrell Software Corp PowerPerl(tm), Add Power To Your Webpage! --------------------------------------------------------------------------- Member of Families Against Internet Censorship: rainbow.rmi.net/~fagin/faic From paulrs at lgs-systems.com Fri Mar 26 06:46:19 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:55 2005 Subject: NSPopUpButton problem - FIX Message-ID: <9903261446.AA10035@slab> Hey guys, Though I received absolutely no responses to my posts on NSPopUpButton (other than complaints about posting example code), I thought someone might actually care that there is a real problem and there is a fix for it. If you recall, I posted an EOF based example which would crash if you used the keyboard to navigate to the pop-up. We were able to narrow the problem down and eliminate the EOF component and found that the issue had to do with calling setTitle on the button with an empty title. Mike Ferris was wonderful enough to take some time and look at the source and responded with this information which I am posting with his permission: _cFlags.mnemonicLocation is not used by NSMenuItemCell and its subclasses. Instead menu item cells ask their menu items for their mnemonic location and return that. This is intentional. When you say setTitle:@"" to a popup, it causes the popup to clear its selected item. A popup with no selected item (ie selectedItem == nil, selectedItemIndex == -1) has a nil menuItem. NSMenuItemCell's implementation of mnemonicLocation was not prepared for that and would return 0 since [nil anyMessage] == 0. You can fix this with a category on NSMenuItemCell that contains the following method: - (unsigned)mnemonicLocation { NSMenuItem *item = [self menuItem]; return (item ? [item mnemonicLocation] : NSNotFound); } This is the new implementation I have put into the kit. Having the category there should not hurt, even when the fix appears. You may want to mark the code, though, so that you can eventually find it and remove it when the kit has the bug fixed for real in a shipping release. The other pop up button issues about which I posted are still unresolved, and I would love to hear from others about them. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From mike at lorax.com Fri Mar 26 08:39:54 1999 From: mike at lorax.com (Mike Ferris) Date: Thu Nov 3 14:41:55 2005 Subject: NSPopUpButton problem - FIX Message-ID: <199903261736.JAA07403@boom.lorax.com> > You can fix this with a category on NSMenuItemCell that contains the following > method: > > - (unsigned)mnemonicLocation { > NSMenuItem *item = [self menuItem]; > return (item ? [item mnemonicLocation] : NSNotFound); > } > Let me add a little caveat to this. Overriding methods in a class with a category is not usually a good idea. There are two potential problems with this. - The behavior of this sort of overriding is technically undefined. If multiple categories and the main class itself implement the same method, you have no real guarantee which will win. The current implementation will provide reproducible behavior for identical cases, but even this is not guaranteed for future releases (hey, the runtime might decide to start calling random() to see who wins :-). In the current implementation a category always wins over the implementation in the main class, but, again, this is not guaranteed to remain true. Be aware, also, that some AppKit classes have their implementations split into multiple categories even within the kit and it is not always possible to know from the headers whether this is the case. Further, this sort of detail is free to change between releases and a method that used to be in the main class implementation may be in a category in the next release. - If you install a category implementation such as the one above and the underlying class' implementation changes dramatically, your category may no longer work properly with the rest of the methods/ivars in the class. That being said, this particular category override is not likely to be too dangerous (although, I am not promising anything). Enough weasel words. Mike Ferris From pierce at calm.twinforces.com Fri Mar 26 10:33:50 1999 From: pierce at calm.twinforces.com (Pierce T. Wetter III) Date: Thu Nov 3 14:41:55 2005 Subject: From the horse's mouth on soundkit Message-ID: <199903261826.LAA07613@calm.twinforces.com> From the about box of OpenAmp, Wilfredo Sanchez's MP3 player: > Note that the NeXT SoundKit is now private API in Mac OS X Server, and will be replaced by > a more useable API in Mac OS X. The output stream here will be rewritten when this > happens, and hopefully the ouput code will be significantly simpler at that point. In > the meantime, it may not be possible to compile this code outside of Apple due to the > lack of published headers for the SoundKit. This may be been already known, but here it is anyways. Pierce From penrose at sfc.keio.ac.jp Fri Mar 26 11:45:57 1999 From: penrose at sfc.keio.ac.jp (Christopher Penrose) Date: Thu Nov 3 14:41:55 2005 Subject: From the horse's mouth on soundkit References: <199903261826.LAA07613@calm.twinforces.com> Message-ID: <36FBE395.869D60AC@sfc.keio.ac.jp> "Pierce T. Wetter III" wrote: > > From the about box of OpenAmp, Wilfredo Sanchez's MP3 player: > > > Note that the NeXT SoundKit is now private API in Mac OS X Server, > and will be replaced by > > a more useable API in Mac OS X. The output stream here will be > rewritten when this > > happens, and hopefully the ouput code will be significantly simpler at > that point. In > > the meantime, it may not be possible to compile this code outside of > Apple due to the > > lack of published headers for the SoundKit. > > This may be been already known, but here it is anyways. > > Pierce Actually the SoundKit is no longer strictly a private framework. The source code for the SoundKit is part of the Darwin open source software distribution. Compile away those melodious MP3 lullabies... http://www.publicsource.apple.com Christopher penrose@sfc.keio.ac.jp From paulrs at lgs-systems.com Fri Mar 26 12:15:52 1999 From: paulrs at lgs-systems.com (Paul R. Summermatter) Date: Thu Nov 3 14:41:55 2005 Subject: Undo confusion Message-ID: <9903262015.AA10520@slab> RE WOF 4.0, NT 4 SP 3 Hey guys, In porting our app to the latest & greatest, I'm having some problems with undo. Specifically, I don't fully understand the documentation on 'Undo and the Responder Chain', and I can find no documentation on the NSWindow delegate method windowWillReturnUndoManager:. Throughout our app, the delegate of the window implements the undo: method and is responsible for sending undo to an EOEditingContext. I have not implemented the method windowWillReturnUndoManager anywhere in our app, and yet, in all but one place, undo works and is sent to the right object. In this one place, our calendar, undo never makes it through to the window's delegate. Can anyone explain what might be going on here? Specifically, how can undo be working if I am not implementing windowWillReturnUndoManager? Further, is there any documentation on windowWillReturnUndoManager? Any information would be greatly appreciated. Regards, Paul --- Paul Summermatter LGS Systems, Inc. Medical Computing Division 15 TJ Gamester Ave Portsmouth, NH 03801-5871 (603) 433-9822 voice (603) 433-9818 fax (888) 898-6321 pager 8986321@skytel.com paging email paulrs@lgs-systems.com (NeXT or MIME Mail Welcome) http://www.lgs-systems.com From thomas.roehl at cai.com Fri Mar 26 12:52:06 1999 From: thomas.roehl at cai.com (Roehl, Thomas) Date: Thu Nov 3 14:41:55 2005 Subject: pthread support in Mac OS X Server Message-ID: <94B9DAE20532D2119FD800805F85CABE01594678@usilmse1.cai.com> Can someone who hasn't gotten their OS X Server already tell me if there is support for the pthread API? Or does it still only provide the cthread thread routines? Thanks in advance From wsanchez at apple.com Fri Mar 26 13:03:31 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:55 2005 Subject: From the horse's mouth on soundkit Message-ID: <199903262103.NAA36538@scv3.apple.com> Hey I'm not the Horse. :-) I don't own SoundKit, but I do know that it is deprecated and that something new will be available for OS X. I imagine that Darwin will want to follow suit, though certainly in the meantime you'll want to use what's available now. That said, if you can use NSSound instead, you should. -Fred From wsanchez at apple.com Fri Mar 26 13:02:33 1999 From: wsanchez at apple.com (Wilfredo Sanchez) Date: Thu Nov 3 14:41:55 2005 Subject: pthread support in Mac OS X Server Message-ID: <199903262102.NAA36950@scv2.apple.com> | Can someone who hasn't gotten their OS X Server already tell | me if there is support for the pthread API? There is no support for pthreads in Mac OS X Server. -Fred From misha at dusya.osd.com Fri Mar 26 13:07:46 1999 From: misha at dusya.osd.com (Misha M. Melikov) Date: Thu Nov 3 14:41:55 2005 Subject: sybase adaptor In-Reply-To: <94B9DAE20532D2119FD800805F85CABE01594678@usilmse1.cai.com> References: <94B9DAE20532D2119FD800805F85CABE01594678@usilmse1.cai.com> Message-ID: <9903262107.AA08950@dusya.osd.com> release notes of macosx state that the sybase eo adaptor will be available SOON. does anybody know how soon is SOON? how about the sybase client libraries for macosx, will they be available? i got the DR2 version of SybaseEOAdaptor.framework to partially work with our app, with the exception of EOModeler not being able to create new models with this adaptor. any input would be appreciated... -m From maury at OAAI.COM Fri Mar 26 13:55:15 1999 From: maury at OAAI.COM (Maury Markowitz) Date: Thu Nov 3 14:41:55 2005 Subject: Getting the measurement units Message-ID: <199903262148.QAA24454@OAAI.COM> Although it's easy enough to get the default units from the userDefaults, there is no published API for getting the measurement units that the user sets in the Page Setup dialog. We punted and used the deaults, but that's not good enough. Does anyone know if there's a way to get this out? I tried looking in the dict after changing it, but I didn't see any entry for this. Maury From fabienr at ncmi.com Fri Mar 26 18:47:44 1999 From: fabienr at ncmi.com (Fabien Roy) Date: Thu Nov 3 14:41:55 2005 Subject: NeXTMail format? In-Reply-To: References: Message-ID: <199903270248.SAA14419@ignem.omnigroup.com> Long time ago I was sending NeXTMail from various UNIXes with the aid of a free script called SendNextMail. I hope that it is still on my old slab. Fabien You wrote: > hi-- > > premise: writing an application that handles NeXTMail. > > is NeXTMail just RTF tarred, compressed, and uuencoded? if not, is > there an actual specification for it? > > and, the next question: is there a facility, either in YB or that > someone else has written, for handling NeXTMail? or would I have to > either roll my own or figure out how to use the Unix utilities? > > any help appreciated. I've checked all the mailing list archives > without success. > > thanks, > chris > > > > ------------------------------- > Chris Doherty > cdoherty@skidmore.edu > > If you cannot find the truth right where you are, where else do you > expect to find it? > -- Dogen > > It is from moments of confusion that we develop moments of clarity. > > --- Fabien L Roy NationsBanc Montgomery Securities, LLC voice: (704)386-75-76 100 North Tryon Street fax: (704)388-95-64 NC1-007-09-08 Charlotte NC 28255 Beeper: 143-9722 (1-800-946-46-46) or http://www.MobileComm.com/message/ email: fabienr@ncmi.com (NeXT/Mime) Pager-email 1439722@mobilecomm.net From theisen at akaMail.com Sat Mar 27 07:30:45 1999 From: theisen at akaMail.com (Dirk Theisen) Date: Thu Nov 3 14:41:55 2005 Subject: Mail parsing code? Message-ID: <1dpc2uu.1hyd0um1ejyk2wM@ascend-tk-p132.rhrz.uni-bonn.de> [ A copy of this message has been posted to the newsgroups ] [ comp.lang.objective-c ] [ gnu.gnustep.discuss ] [ comp.sys.next.programmer ] Hi! Has anyone some nice ObjC mail (RFC822) and/or MIME (RFC1521) parsing code lying around that he'd like to share for a non-commercial project? Regards, Dirk [who doesn't like to reinvent the wheel] -- Buy a Pentium III now, get your personal Big Brother FREE!!! http://theisen.home.pages.de/ From yami at dementian.com Sat Mar 27 09:14:19 1999 From: yami at dementian.com (Steven D. Arnold) Date: Thu Nov 3 14:41:55 2005 Subject: number class Message-ID: <199903271755.MAA12558@phear.dementian.com> Hi, I am looking for a number class in Obj-C that has the following features: - Handles arbitrarily large (within limits of RAM) floating-point numbers - Provides accuracy to a selected degree - Includes most important mathematical functions -- the basic operators of course, but also powers, logarithms, trig - Capable of representing itself as a string in the following formats: - decimal number - exponential/scientific/engineering notation - hexadecimal, binary, octal - Capable of representing itself as a fraction instead of a decimal number - Features such as choosing the number of decimal places to display in the string representation. The more of these things a class can do, the better: I can subclass and add features if not all these are implemented. Is NSDecimalNumber my best choice? I hope there's something nicer. steve From frank at inis.de Sun Mar 28 09:44:02 1999 From: frank at inis.de (Frank Knobloch) Date: Thu Nov 3 14:41:55 2005 Subject: YB under NT - Printing to a Network-Printer Message-ID: <199903281744.SAA00566@inis-2.inis.de> Hi all, are there any ideas on how to print on a non-ps network-printer that has it's own printbox (Jet-Direct). It's a normal line-printer and we only want to print some ascii-lines on it. If it's connected to a serial port, it's no problem. Any hints? ------------------- //InIS GmbH * Loesungen fuer Automations- und Verwaltungssysteme [Addr.: Friedrich-Ebert-Str. 78 ~ D-34119 Kassel] [Tel..: ++49-561-788090, Fax: ++49-561-78 80 929] [Mail.: F.Knobloch@inis.de, Web: www.inis.de] //Never trust existing sources! From frank at inis.de Sun Mar 28 09:45:11 1999 From: frank at inis.de (Frank Knobloch) Date: Thu Nov 3 14:41:55 2005 Subject: Yellow Box NT - CD Applications Message-ID: <199903281745.SAA00571@inis-2.inis.de> Has anybody figured out how it will be possible to developer applications with YB under NT, running from a CD-ROM. I think the only problem will be the services. ------------------- //InIS GmbH * Loesungen fuer Automations- und Verwaltungssysteme [Addr.: Friedrich-Ebert-Str. 78 ~ D-34119 Kassel] [Tel..: ++49-561-788090, Fax: ++49-561-78 80 929] [Mail.: F.Knobloch@inis.de, Web: www.inis.de] //Never trust existing sources! From frank at inis.de Sun Mar 28 11:13:55 1999 From: frank at inis.de (Frank Knobloch) Date: Thu Nov 3 14:41:55 2005 Subject: Really interesting code... Message-ID: <199903281913.UAA00633@inis-2.inis.de> SENSATION: OPEN SOURCE CODE OF WIN98 !!! Object: code source windows 98 Date: mercredi 8 juillet 1998 13:43 /* TOP SECRET Microsoft Code Project: Chicago (TM) Projectet Release(c)date: Summer 1994 */ #include "win31.h" #include "win95.h" #include "evenmore.h" #include "oldstuff.h" #include "billrulz.h" #define INSTALL = HARD char_make_prog_look_big[0xffffffff]; void main() { while(!CRASHED) { display_copyright_message(); display_bill_rules_message(); do_nothing_loop(); if (first_time_installation) { make_50_megabytes_swapfile(); do_nothing_loop(); totally_screw_up_HPFS_file_system(); search_and_destroy_the_rest_of_OS/2(); hang_system(); uninstall_netscape_applications(); } write_something(anything); disply_copyright_message(); do_nothing_loop(); void create_useless_network_traffic(); do_some_stuff(); if (still_not_crashed) { display_copyright_message(); do_nothing_loop(); basically_run_windows_31(); do_nothing_loop(); if (netscape_applications_reinstalled) { make_ie4_primary_browser(); gradually_disable_netscape_applications(all); } else { make_disks_spin_to_make_look_like_a_real_os_like_unix() ; do_nothing_loop(); } } } if (detect_cache()) { disable_cache(); } if (too_many_applications_running) { start_crash_count_down(); begin_manadatory_file_corruption(); /* set_chdsk_flag(); */ set_scan_disk_flag(); disable_logins(); } if (fast_cpu) { set_wait_states(lots); set_mouse(speed, very_slow); set_mouse(action, jumpy); set_mouse(reaction, sometimes); } /* printf("Welcome to Windows 3.11"); */ /* printf("Welcome to Windows 95"); */ printf("Welcome to Windows 98"); if (system_ok()) { crash(to_dos_prompt); } else { system_memory = open("a:\swp0001.swp", O_CREATE); while(something) { sleep(5); get_user_input(); sleep(5); act_on_user_input(); sleep(5); if (novell_found) { mk_ipx_more_noisy(); make_new_connections_fail_randomly(); disconnect_printers(); attempt_to_remotely_crash_netware_server(); sleep(60); launch_message("You should switch to Windows NT"); } if (! exchange_found) { prompt_repeatedly_for_user_to_install_exchange(); } if (connected_to_internet) { email_pc_configuration_to_bill(); email_list_of_installed_applications_to_bill(); send_list_of_illegal_apps_to_fbi(); send_turbo_tax_data_files_to_irs(); } } if (keypressed() || netscape_apps_installed || novell_apps_installed) { reboot(); } else { create_general_protection_fault(); } } } ------------------- //InIS GmbH * Loesungen fuer Automations- und Verwaltungssysteme [Addr.: Friedrich-Ebert-Str. 78 ~ D-34119 Kassel] [Tel..: ++49-561-788090, Fax: ++49-561-78 80 929] [Mail.: F.Knobloch@inis.de, Web: www.inis.de] //Never trust existing sources! From karl at matthias.org Sun Mar 28 14:24:40 1999 From: karl at matthias.org (Karl Matthias) Date: Thu Nov 3 14:41:55 2005 Subject: __objcInit error Message-ID: <199903282224.RAA21784@matthias.org> I just installed MacOS X Server (I'm a former NeXT fanatic) and have been hacking with it all weekend. I found some things missing which I use very often on OpenBSD, so I decided to port them over to OS X. I got and compiled the OpenBSD versin of ksh and it now runs fine. I then attempted to compile the OpenBSD version of ftp which supports both ftp and http URLs on the command line. After some hacking it compiles, runs, and http URLs work fine, but I get the following error once I have logged in to an ftp site: dyld: ./ftp multiple definitions of symbol __objcInit ./ftp definition of __objcInit /System/Library/Frameworks/System.framework/Versions/B/System(objc-runtime.o) definition of __objcInit Anyone have ANY idea how to get rid of that? Thanks for your help! Karl N. Matthias P.S. What follows is the complete ftp session: $ ./ftp ftp://ftp.ou.edu/pub/ Connected to atlas.services.ou.edu. 220 atlas.services.ou.edu FTP server (NcFTPd 2.3.1, free educational license) ready. 331 Guest login ok, send your complete e-mail address as password. 230-You are user #214 of 300 simultaneous users allowed. 230- 230 Logged in anonymously. Remote system type is UNIX. Using binary mode to transfer files. 200 Type okay. 250 "/" is new cwd. 250 "/pub" is new cwd. dyld: ./ftp multiple definitions of symbol __objcInit ./ftp definition of __objcInit /System/Library/Frameworks/System.framework/Versions/B/System(objc-runtime.o) definition of __objcInit From sanguish at digifix.com Sun Mar 28 20:06:03 1999 From: sanguish at digifix.com (Scott Anguish) Date: Thu Nov 3 14:41:55 2005 Subject: YB under NT - Printing to a Network-Printer In-Reply-To: <199903281744.SAA00566@inis-2.inis.de> References: <199903281744.SAA00566@inis-2.inis.de> Message-ID: <199903290406.XAA28918@digifix.com> Frank Knobloch wrote: > Hi all, > > are there any ideas on how to print on a non-ps network-printer > that has it's own printbox (Jet-Direct). > It's a normal line-printer and we only want to print some ascii-lines > on it. If it's connected to a serial port, it's no problem. > Any hints? Better than hints.. how about a step-by-step instruction? Wassim Jabi wrote up his techniques for doing this for Stepwise, and its currently on the front page (http://www.stepwise.com/) Hows that for timing? :-) From stachura at apple.com Sun Mar 28 22:59:41 1999 From: stachura at apple.com (Renata Stachura Poray) Date: Thu Nov 3 14:41:55 2005 Subject: *Comprehensive OpenStep developer training April, 1999 in Reston, VA* References: <36FC4E1E.7D0EF5ED@apple.com> <36FC6AF9.2C68147@apple.com> <36FDE612.FCD5EFDC@apple.com> <36FEF748.75C95938@apple.com> Message-ID: <36FF24DD.4A686FAA@apple.com> Apple Enterprise Training invites OpenStep programmers to attend one or all of our series of OpenStep classes in Reston, VA in April. Programming OpenStep I --- April 6-9 Programming OpenStep II --- April 12-16 Enterprise Objects for OpenStep --- April 26-30 To enroll please call Apple Enterprise at (800) 879-6398 or your local Apple Enterprise sales rep. Seating is limited. For more detailed course information and the on-line schedule, please see our website: http://www.apple.com/enterprise/training/schedule.html Sincerely, Renata Poray Apple Enterprise Training Manager -------------- next part -------------- An HTML attachment was scrubbed... URL: /mailman/archive/macosx-dev/attachments/19990328/ed9a514b/attachment.html From cld at europay.telekurs.com Sun Mar 28 23:58:00 1999 From: cld at europay.telekurs.com (Daniel Clausen) Date: Thu Nov 3 14:41:55 2005 Subject: Really interesting code... References: <199903281913.UAA00633@inis-2.inis.de> Message-ID: <36FF3288.33841AD@europay.telekurs.com> Hi, frank@inis.de wrote: > > SENSATION: OPEN SOURCE CODE OF WIN98 !!! [snip] > void main() [snip] I dunno about the rest, but ANSI-C forbids void main(). :-) cu, -sargon -- Happiness lies in being privileged to work hard for long hours in doing whatever you think is worth doing. -- Jubal Harshaw, To Sail Beyond the Sunset From erik at object-factory.com Mon Mar 29 02:44:20 1999 From: erik at object-factory.com (Erik Doernenburg) Date: Thu Nov 3 14:41:55 2005 Subject: Mail parsing code? Message-ID: <199903291044.CAA20598@ignem.omnigroup.com> > Has anyone some nice ObjC mail (RFC822) and/or MIME (RFC1521) parsing > code lying around that he'd like to share for a non-commercial project? Yes. I'm currently working on ALX3000 (the Mac OS X version of the NEXTSTEP newsreader Alexandra) and the Message handling subproject is almost finished. If you can wait for a week or so, which I will need to finalize some design issues, you can get access to the code via our public FTP server and/or CVS. Note, however, that ALX3000 will be released under GPL2 and you will have to accept it if you decide to use the code. Most of the implementation already works well and I'd be extremely pleased if you use this code as it not trivial to do an implementation that is compliant to the corresponding standards and still accepts most of the junk that many clients generate. The latter can only be ensured if many people check the code. regards, erik -- OBJECT FACTORY, Gesellschaft fuer Informatik und Datenverarbeitung mbH Telephone: ++49 +231 9751370 . Internet: http://www.object-factory.com From janos.lobb at yale.edu Mon Mar 29 07:19:07 1999 From: janos.lobb at yale.edu (=?iso-8859-1?Q?J=E1nos?= =?iso-8859-1?Q?_?= =?iso-8859-1?Q?L=F6bb?=) Date: Thu Nov 3 14:41:55 2005 Subject: Really interesting code... In-Reply-To: <36FF3288.33841AD@europay.telekurs.com> Message-ID: At 3:08 -0500 3/29/1999, Daniel Clausen wrote: >Hi, > >frank@inis.de wrote: >> >> SENSATION: OPEN SOURCE CODE OF WIN98 !!! >[snip] >> void main() >[snip] > >I dunno about the rest, but ANSI-C forbids void main(). > >:-) > >cu, > -sargon > >-- >Happiness lies in being privileged to work hard for >long hours in doing whatever you think is worth doing. > -- Jubal Harshaw, To Sail Beyond the Sunset And it does not follow Mr. Simonyi's Hungarian notation -:) J?nos J?nos L?bb Tel: 203-737-5204 Yale University Pathology Fax: 203-785-7303 310 Cedar St. Room BML104A janos.lobb@yale.edu New Haven CT 06510 Never take a candid-cookie from a stranger. From dsolis at data.net.mx Mon Mar 29 10:58:14 1999 From: dsolis at data.net.mx (David Solis) Date: Thu Nov 3 14:41:55 2005 Subject: NSBox subclass on IBPalette - adding subviews in IB Message-ID: <9903291858.AA00958@data.net.mx> Hi Andreas, Some time ago I has the same problem. My solution was: 1. In your palette, add a instance variable to your subclase of IBPalette, for instance snapView. 2. In IB, connect the variable to the widget . 3. In the method finishInstantiate add the following code: [[self paletteDocument] attachObject:[snapView contentView] toParent:snapView]; if you've had success with this,please let me know. Regards Dd -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/enriched Size: 464 bytes Desc: not available Url : /mailman/archive/macosx-dev/attachments/19990329/2df42aec/attachment.bin From karl at matthias.org Mon Mar 29 18:28:17 1999 From: karl at matthias.org (Karl Matthias) Date: Thu Nov 3 14:41:55 2005 Subject: more __objcInit Error Message-ID: <199903300227.VAA24045@matthias.org> Thanks to David Young... but I still haven't got this one licked. ANY other suggestions are very much appreciated. If someone would be willing to run the compile on their machine (OS X Server 1.0) to see if they don't have the same error, I would also be very appreciative(mail me if you want to!). Here's what I've done: Checked all my .o files for the objcInit symbol Attempted to statically link the binary (didn't have the right options? didn't work) Checked the headers in all the files in an attempt to see which libraries were called Attempted to poke around with gdb to find where the symbol was coming from(not enough gdb experience, apparently) I noticed one other strange error. When I compile the code with db.h listed AFTER dirent.h in the code, dirent.h produces errors saying: /System/Library/Frameworks/System.framework/Headers/bsd/sys/dirent.h:29: undefined type, found `u_int32_t' /System/Library/Frameworks/System.framework/Headers/bsd/sys/dirent.h:30: undefined type, found `u_int16_t' /System/Library/Frameworks/System.framework/Headers/bsd/sys/dirent.h:31: undefined type, found `u_int8_t' /System/Library/Frameworks/System.framework/Headers/bsd/sys/dirent.h:32: undefined type, found `u_int8_t' This seems rather wrong to me, given that this is a system header file... It goes away if my code includes db.h beforehand, so I'm wondering if anyone thinks this might be the cause somehow? Thanks for the help! Karl From Joe.Loda at usmail.mpct.com Tue Mar 30 12:26:02 1999 From: Joe.Loda at usmail.mpct.com (Loda, Joe) Date: Thu Nov 3 14:41:55 2005 Subject: Replacement for Terminal.app on NT Message-ID: As we investigate development on NT with WO4, all of us mach developers are realizing just how helpful Terminal.app is. It appears that there is no effort being made to replace it. With an NT command window, cutting and pasting is limited, and searching is out of the question. Does anyone have any suggestions for a replacement? I've tried TakeCommand/32 from JPSOFT, but the pasting doesn't work right, and there is still no search capabilities. I wish they'd just release Terminal.app to the Darwin project or MiscKit. Any ideas? Thanks, Joe --- Joe Loda, mpct Solutions Corporation, Chicago joe.loda@mpct.com (ASCII, MIME) From Joe.Loda at usmail.mpct.com Tue Mar 30 12:40:08 1999 From: Joe.Loda at usmail.mpct.com (Loda, Joe) Date: Thu Nov 3 14:41:55 2005 Subject: Terminal.app Message-ID: I should have mentioned - Shell.app on NT is not a suitable replacement (at least, as is), as it's handling of control-c has problems. When we were using it, sometimes control-c would "re-awaken" the shell rather than just propagating the break to gdb. Then, it seemed like both gdb and the shell were "fighting" for stdin. When NeXT/Apple support was contacted regarding this, they said "Yup. Don't do that". Joe --- Joe Loda, mpct Solutions Corporation, Chicago joe.loda@mpct.com (ASCII, MIME) From zmonster at heliosgroup.com Tue Mar 30 12:52:51 1999 From: zmonster at heliosgroup.com (Eric Hermanson) Date: Thu Nov 3 14:41:55 2005 Subject: Replacement for Terminal.app on NT Message-ID: <5948F7268092D211B2020000D11C0FB80817B7@chi-serv2.heliosgroup.com> > -----Original Message----- > From: Loda, Joe [mailto:Joe.Loda@usmail.mpct.com] > As we investigate development on NT with WO4, all of us mach > developers are realizing just how helpful Terminal.app is. It appears that > there is no effort being made to replace it. With an NT command window, > cutting and pasting is limited, and searching is out of the question. > > Does anyone have any suggestions for a replacement? The best one I've found is CRT from VanDyke - it's the best $30 you'll ever spend for shareware. http://www.vandyke.com Eric From ttakeo at amtec.co.jp Tue Mar 30 21:16:56 1999 From: ttakeo at amtec.co.jp (tetuya takeo) Date: Thu Nov 3 14:41:55 2005 Subject: Apple's WWDC, there are no YB. Message-ID: <9903310508.AA13272@amtec000.amtec.co.jp> YB Developers, I checked Apple's WWDC home page, but there are no mentions about YellowBox, and lecturers are very different from previous years. I wonder this year's WWDC deserves to attend for yellow developer like me. Am I missing important information, or totally misunderstanding? Like to hear your opinions. Regards, --- Tetuya TAKEO ttakeo@amtec.co.jp (NeXTmail/MIME accepted) From aswift at redmud.com Wed Mar 31 09:17:05 1999 From: aswift at redmud.com (Adam Swift) Date: Thu Nov 3 14:41:55 2005 Subject: Omni CVS/EOModeler bundle Message-ID: <37025895.FA0DE5F6@redmud.com> Hello, I've tried installing the OmniCVSEOModelerBundle and I can't seem to get it configured under MacOSX. I installed the bundle in "/Network/Developer/EOMBundles" (which appears in the EOModeler bundle search path list) and wrote out the default to load it as instructed: defaults write EOModeler BundlesToLoad /Network/Developer/EOMBundles/OmniCVSEOModelerBundle.bundle But every time I save my model it deletes the CVS directory from the .eomodeld directory. Any clues? Does anyone have this working under MacOSX? Thanks, - adam From andrew at omnigroup.com Wed Mar 31 20:12:11 1999 From: andrew at omnigroup.com (Andrew Abernathy) Date: Thu Nov 3 14:41:55 2005 Subject: Omni CVS/EOModeler bundle Message-ID: <199904010412.UAA07406@ignem.omnigroup.com> > I've tried installing the OmniCVSEOModelerBundle and I can't seem to get > it configured under MacOSX. [...] > But every time I save my model it deletes the CVS directory from the > .eomodeld directory. EOModeler for EOF 3.0 (which is what comes with MXS) deals with bundles differently. Also, a few changes have been made to the bundle. (A bug fix dealing with deleting then re-adding entities and stored procedures; addition of support for prebuilt fetch specifications; and renaming the extension to .EOMbundle so EOModeler likes it. I don't think anything else was changed.) The current bundle _might_ work on MXS if you renamed its extension from .bundle to .EOMbundle (and made sure it was in an appropriate directory, as specified by EOModeler preferences - it looks like your current directory is correct). However, I'll make sure to get an updated bundle out tonight. (I'll send a notice to omninews and make an entry on our news web page when it's done.) -andrew From Philippe.Robert at uptime.ch Wed Mar 31 23:59:19 1999 From: Philippe.Robert at uptime.ch (Philippe C.D. Robert) Date: Thu Nov 3 14:41:55 2005 Subject: PDO and SUN In-Reply-To: <199904010830.AA00484@proxyware.fr> References: <199904010830.AA00484@proxyware.fr> Message-ID: <9904010759.AA00436@uptime.ch> You wrote: > I mean we need FoundationKit to run on SUN/Solaris (not our OS choice!) If you just need 'a' FoundationKit, you can use gnustep-base, it is nearly complete and very usable! sweet dreams, Phil --- Philippe C.D. Robert Uptime Object Factory Inc http://www.nice.ch/~phip Software Development http://www.nice.ch/ProjectCenter OpenStep/Unix