Re: Chat with Thomas Dybdahl Ahle
Thomas Dybdahl Ahle
> Hi Thomas,
>
> On Sat, 14 Apr, 2007 at 11:26:26PM +0200, Thomas Dybdahl Ahle wrote:
> > l??r, 14 04 2007 kl. 23:10 +0530, skrev Varun Hiremath:
> > > Please find attached the deb package for the final version.
> > Thanks.
> >
> > > But, sorry to say that it still freezes on my comp :-(
> > Dammit. Don't you have a file called /etc/debian-release
> > or /etc/debian_version?
>
> I have /etc/debian_version file.
ok. Then exporting manually should make do difference.
> > Could you try to run export LD_ASSUME_KERNEL=2.4.1 just before you start
> > pychess?
>
> Actually yesterday I upgraded my computer to the latest Debian
> unstable. Now pychess refuses to start and gives this error. Before
> this upgrade pychess final version didn't give this error.
>
> $$ python: error while loading shared libraries: libpthread.so.0: cannot
> open shared object file: No such file or directory
>
> I changed the export LD_ASSUME_KERNEL to 2.6.1 in pychess file and
> then the main window loaded but still it freezes when I start playing
> the game.
>
> What exactly does this export statement do? Is the problem with the
> new kernels or the old kernels?
I seams to be related to some pthreads 2.4->2.6 stuff.
http://www.daa.com.au/pipermail/pygtk/2005-October/011246.html
I get the same error when using the line on fedora, so probably debian
has now switched to theo same kernel configuration on the point.
I guess we should the remove the auto calling on debian, and instead
write it in some "known errors" article...
What I just don't seam to understand is why pychess still crashes on you
computer...
idle_add and source_remove seams to be a common reason for crashes:
http://www.google.com/search?q=idle_add+freeze but everybody else seams
to solve it by using ASSUME_KERNEL on debian.
Perhaps we could recode PyChess to not use source_remove. I might not be
too hard. Not sure.
--
Med venlig hilsen,
Best regards,
Thomas
Ondrej Certik
ondra@syslik:~/pychess$ ./pychess
Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/pychess/Main.py", line 138,
in on_new_game1_activate
ionest.newGame ()
File "/usr/lib/python2.4/site-packages/pychess/widgets/ionest.py",
line 258, in newGame
game, gmwidg = runNewGameDialog()
File "/usr/lib/python2.4/site-packages/pychess/widgets/ionest.py",
line 234, in runNewGameDialog
engine0, args0 = engine1, args1 = anaengines[0]
IndexError: list index out of range
I just run pychess and a "new game", but it stopped responding, I
couldn't move pieces, nothing.
Ondrej
On Apr 15, 12:20 pm, Thomas Dybdahl Ahle <lob...@gmail.com> wrote:
> søn, 15 04 2007 kl. 15:31 +0530, skrev Varun Hiremath:
>
> > Hi Thomas,
>
> > On Sat, 14 Apr, 2007 at 11:26:26PM +0200, Thomas Dybdahl Ahle wrote:
> > > l??r, 14 04 2007 kl. 23:10 +0530, skrev Varun Hiremath:
> > > > Please find attached the deb package for the final version.
> > > Thanks.
>
> > > > But, sorry to say that it still freezes on my comp :-(
> > > Dammit. Don't you have a file called /etc/debian-release
> > > or /etc/debian_version?
>
> > I have /etc/debian_version file.
>
> ok. Then exporting manually should make do difference.
>
>
>
> > > Could you try to run export LD_ASSUME_KERNEL=2.4.1 just before you start
> > > pychess?
>
> > Actually yesterday I upgraded my computer to the latest Debian
> > unstable. Now pychess refuses to start and gives this error. Before
> > this upgrade pychess final version didn't give this error.
>
> > $$ python: error while loading shared libraries: libpthread.so.0: cannot
> > open shared object file: No such file or directory
>
> > I changed the export LD_ASSUME_KERNEL to 2.6.1 in pychess file and
> > then the main window loaded but still it freezes when I start playing
> > the game.
>
> > What exactly does this export statement do? Is the problem with the
> > new kernels or the old kernels?
>
> I seams to be related to some pthreads 2.4->2.6 stuff.http://www.daa.com.au/pipermail/pygtk/2005-October/011246.html
>
> I get the same error when using the line on fedora, so probably debian
> has now switched to theo same kernel configuration on the point.
>
> I guess we should the remove the auto calling on debian, and instead
> write it in some "known errors" article...
>
> What I just don't seam to understand is why pychess still crashes on you
> computer...
> idle_add and source_remove seams to be a common reason for crashes:http://www.google.com/search?q=idle_add+freezebut everybody else seams
Thomas Dybdahl Ahle
1) PyChess engine got timeout.
2) PyChess engine crashed.
If the latest one is true, you should be able to get a more interesting
message from your log file.
If the first one is true, you shouldn't see the problem very often, and
in 0.7 it should be completely fixed due to much better engine handling
code.
It might be relevant however to come up with a better errormessage..
Thomas Dybdahl Ahle
> On Thu, 19 Apr, 2007 at 09:36:33PM +0200, Thomas Dybdahl Ahle wrote:
> > fre, 20 04 2007 kl. 00:23 +0530, skrev Varun Hiremath:
> > > Hi Thomas,
> > > http://www.async.com.br/faq/pygtk/index.py?req=show&file=faq03.007.htp
> >
> > Yeah I know, but no hard tasks are performed in a callback as far as I
> > remeber.
> > Did you have 100% usage on the freezes?
>
> I didn't get you. What do you mean by 100% usage?
I mean, that the problem solved by not using source_remove which cursed
people like Ondrej Certik, I believe PyChess didn't use 100% of the cpu.
This means that your problem might be different and coursed by some
infinite loop in a callback...
Problem is that it is damn hard to trace.
Perhaps using a profiler would help. The problem is that the python
profilers I know of only print their results when the code they are
profiling exits, what a infinite loop (or similar) by definition never
will.
Also if the problem lay in code that way, it is strange that nobody else
sees.
But on the other hand, if you have just updated your entire system, it
is hard to imagine the problem being in your libraries..