OW 5.1 Final Released

[Jake Robb]

The only information I have right now that might be helpful is that when you
close OW, normally it takes a moment (possibly not a noticeable one on your
G5?), which I assume is to write the active workspace's data to the file.
If it takes this moment, you're OK.  But once in a while, OW exits very
quickly.  On the next launch, all of the windows in that workspace will be
gone!

Based on that, I would suggest looking for possible points of failure early
in the workspace file writing process triggered by a Quit.

I will attempt to gather more information -- I hope what I have given is
useful.  If some logging and debugging code can be added, I would gladly run
that version.

-Jake


Ken Case wrote:

> On Jan 6, 2005, at 6:21 AM, Jake Robb wrote:
>> So am I correct to assume that the problem where OW "forgets" about
>> all of your workspaces still exists?
> 
> Apparently it does:  when I saw the reports on the list yesterday I
> reopened bug #16421, but this was after we had frozen 5.1 for release.
> We didn't unfreeze 5.1 because this problem isn't a regression from
> 5.0--that is, it isn't new to 5.1--and we thought it was more important
> to get 5.1's other fixes (with the new WebCore, better caching, and
> bookmark synchronization fixes) out to all our customers.
> 
> (Also, most of our users don't seem to encounter this problem.  I know
> that I use Workspaces a _lot_ and I've never lost one.)
> 
> This is our highest priority bug, however, and I'd love to collect more
> information so we can solve it.
> 
>> As intermittent as it is, I can see how it would be quite difficult to
>> diagnose and debug, but nonetheless, it is a show-stopper for me.
> 
> It would be for me, too!  I contacted Piers yesterday to collect more
> information from him, but any information you could send me would be
> great, too.
> 
> What I most need to know is when the workspace information is lost:  in
> particular, is it lost when writing the file, or when reading it?  (On
> quit, or on launch?)  In an ideal world, I'd like a look at the
> workspace file just before it gets corrupt and just after, and exactly
> what happened between those states.
> 
> If it's hard to collct that information manually, I can try adding some
> debugging tools to our next beta to collect that information (by saving
> off copies and writing to a log file each time it writes and reads).
> 
> Ken
>