2.8.12

[chris elliott]

Julian and others: are there any reasons not to roll 2.8.12 this evening ?

chris

[Julian Smart]

Hi Chris,

On 14/03/2011 08:56, chris elliott wrote:
> Julian and others: are there any reasons not to roll 2.8.12 this
> evening ?

Probably no reason why not; I'm investigating a report from one user of
a timer-related crash on exit that I wasn't getting when using wxMSW
2.8.10 which has a different timer implementation, but since I haven't
seen any other indications of this I have to assume this is app-specific.

I wonder if someone could test binary backward compatibility, especially
for wxGTK, before we release 2.8.12?

Thanks,

Julian

>
> chris
>

--
Julian Smart, Anthemion Software Ltd.
www.anthemion.co.uk | +44 (0)131 229 5306
Tools for writers: www.writerscafe.co.uk
Ebook creation: www.jutoh.com
wxWidgets RAD: www.dialogblocks.com

[Vadim Zeitlin]

On Mon, 14 Mar 2011 09:20:01 +0000 Julian Smart <jul...@anthemion.co.uk> wrote:

JS> I wonder if someone could test binary backward compatibility, especially
JS> for wxGTK, before we release 2.8.12?

I don't think we have any automatic way to do this so the best I can think
of is to build a biggish wx application (such as DialogBlocks for example)
with 2.8.0 linking it dynamically and then run it with 2.8.12. If this
works, chances are that we didn't do anything too stupid.

Of course, it would be nice to have some more fool proof way of ensuring
this. I don't know of any good tool to do it. A quick search found a couple
of them: a very simple one at https://github.com/glk/shlib-compat and
another one at http://ispras.linux-foundation.org/index.php/ABI_compliance_checker
but I don't have any personal experience with them, does anybody?

Regards,
VZ

[Chauk-Mean]

Hi,

On 14 mar, 10:20, Julian Smart <jul...@anthemion.co.uk> wrote:
> Hi Chris,
>
> On 14/03/2011 08:56, chris elliott wrote:> Julian and others: are there any reasons not to roll2.8.12this
> > evening ?
>
> Probably no reason why not; I'm investigating a report from one user of
> a timer-related crash on exit that I wasn't getting when using wxMSW
> 2.8.10 which has a different timer implementation, but since I haven't
> seen any other indications of this I have to assume this is app-specific.

I've also experienced the same issue on MingW with 2.8.11 and 2.8.12-
rc2.
This is a wxRuby application that makes use of a timer for animating
graphics.
The application works fine with wxWidgets 2.8.10.
But with 2.8.11 and 2.8.12-rc2, it crashes at the application exit.

Thanks.
Chauk-Mean

[Julian Smart]

Hi,

On 16/03/2011 07:22, Chauk-Mean wrote:
>
>> Probably no reason why not; I'm investigating a report from one user of
>> a timer-related crash on exit that I wasn't getting when using wxMSW
>> 2.8.10 which has a different timer implementation, but since I haven't
>> seen any other indications of this I have to assume this is app-specific.
> I've also experienced the same issue on MingW with 2.8.11 and 2.8.12-
> rc2.
> This is a wxRuby application that makes use of a timer for animating
> graphics.
> The application works fine with wxWidgets 2.8.10.
> But with 2.8.11 and 2.8.12-rc2, it crashes at the application exit.

That's quite worrying; does it happen every time, and did you stop the
timer before closing the app? Did you manage to run the app in a
debugger to confirm where it was crashing?

I wonder if we should add a function to timer.cpp to set a variable for
ignoring all timer messages, and call this function from wxApp::CleanUp.
However this doesn't really address the root cause of the problem.

I haven't been able to reproduce the crash myself which is frustrating,
and I've confirmed that all timers are stopped before exiting and (on my
XP system at least) Notify isn't being called after Stop is called.

Any ideas on this appreciated. Perhaps we'd better suspend the 2.8.12
release for a little while.

Regards,

Julian
> Thanks.
> Chauk-Mean

[Chauk-Mean]

Hi Julian,

On 16 mar, 10:20, Julian Smart <jul...@anthemion.co.uk> wrote:

> >> Probably no reason why not; I'm investigating a report from one user of
> >> a timer-related crash on exit that I wasn't getting when using wxMSW
> >> 2.8.10 which has a different timer implementation, but since I haven't
> >> seen any other indications of this I have to assume this is app-specific.
> > I've also experienced the same issue on MingW with 2.8.11 and 2.8.12-
> > rc2.
> > This is a wxRuby application that makes use of a timer for animating
> > graphics.
> > The application works fine with wxWidgets 2.8.10.
> > But with 2.8.11 and 2.8.12-rc2, it crashes at the application exit.
>
> That's quite worrying; does it happen every time, and did you stop the
> timer before closing the app? Did you manage to run the app in a
> debugger to confirm where it was crashing?

Sorry, the crash is related with the paint event (for drawing
graphics) and not with the timer.
I left the timer and commented the paint event and the crash does not
occur anymore.
I've also tried a sample with OpenGL animation relying on a timer and
everything works fine.

Are there some significant changes in Graphics and/or GraphicsContext
between 2.8.10 and 2.8.11+ ?

> Perhaps we'd better suspend the 2.8.12
> release for a little while.

As the crash does not seem to be related with timer, I don't think
it's worth suspending the 2.8.12 release.

Thanks.
Chauk-Mean.

[Julian Smart]

Hello Chauk-Mean,

On 16/03/2011 14:04, Chauk-Mean wrote:
>
> Sorry, the crash is related with the paint event (for drawing
> graphics) and not with the timer.
> I left the timer and commented the paint event and the crash does not
> occur anymore.
> I've also tried a sample with OpenGL animation relying on a timer and
> everything works fine.

That's sort of good news, thanks.

> Are there some significant changes in Graphics and/or GraphicsContext
> between 2.8.10 and 2.8.11+ ?

I don't know, I'm afraid. Perhaps you can check in gdb where the crash
is happening exactly.

>
> As the crash does not seem to be related with timer, I don't think
> it's worth suspending the 2.8.12 release.

Since no-one else has reported it, I think you're right, we're OK for a
2.8.12 release (thanks Chris - whenever you have time).

Best regards,

Julian
> Thanks.
> Chauk-Mean.

[Angel Vidal]

  I know it's a bit late, but could I have #13055 considered for 2.8.12 if it's not packaged yet? It fixes a nasty random crash in wxGTK (accessing an already freed gsocket object) when using wxSocket inside threads. For example, aMule is affected by this crash and a pthread version of the patch attached to the ticket was confirmed to fix it for a user that was affected by it in a way that would crash the application very shortly after running it. After a suggestion by Vadim, I changed the patch to use wxMutex instead.  I have also confirmed locally that the patch fixes #10178, which contains a code example that used to crash in my debian box after just a couple presses of the F1 key to create threads that use sockets in them. 

  All in all, I think it'd be a safe and good addition to 2.8.12 if it's still possible. I was planning on finishing this way earlier, but events beyond my control conspired to prevent me from it.

  Thanks,

  Angel Vidal (Kry)

[Julian Smart]

Hi Angel,

On 19/03/2011 19:39, Angel Vidal wrote:
> I know it's a bit late, but could I have**#13055 considered for

I think it's worth the risk and the socket sample is working fine with
the patch, so I've committed it. AFAIK the 2.8.12 release hasn't been
spun yet.

Thanks!

Julian
>
> Thanks,
>
> Angel Vidal (Kry)
> --
> To unsubscribe, send email to wx-dev+un...@googlegroups.com
> or visit http://groups.google.com/group/wx-dev

[Angel Vidal]

  On Sat, Mar 19, 2011 at 3:31 PM, Julian Smart <jul...@anthemion.co.uk> wrote:

Hi Angel,

I think it's worth the risk and the socket sample is working fine with the patch, so I've committed it. AFAIK the 2.8.12 release hasn't been spun yet.

  Good! And good catch regarding the missing #include, sorry for the oversight.

  Now, as for the ABI discussion above, I decided to try abi-compliance-checker on 2.8.11 vs branch 2.8 and 2.8.0 vs branch 2.8. I had to patch 2.8.0 with http://trac.wxwidgets.org/changeset/61009 or it wouldn't compile, but that's a trivial change.

  I compiled with --enable-unicode and --enable-debug, and I'm happy to report that both the headers and shared objects tests were successful, both against 2.8.0 and 2.8.11. 

  I tried with --disable-debug, and it reported success as well, but showed some warnings about problems compiling the headers (for all versions). The .so files compared fine. 

  Let me know if you want me to test any other linux configurations. 

  abi-compilance-checker generates HTML reports, I'll be happy to attach them or upload them to my server if you want to see them, but they basically list the headers and libraries and the compatibility problems, and those are all 0 for all tests. 

  Regards, 

  Angel Vidal (Kry)

[Julian Smart]

Hi Angel,

On 19/03/2011 21:10, Angel Vidal wrote:
> On Sat, Mar 19, 2011 at 3:31 PM, Julian Smart
> <jul...@anthemion.co.uk <mailto:jul...@anthemion.co.uk>> wrote:
>
> Hi Angel,
>
> I think it's worth the risk and the socket sample is working fine
> with the patch, so I've committed it. AFAIK the 2.8.12 release
> hasn't been spun yet.
>
>
> Good! And good catch regarding the missing #include, sorry for the
> oversight.

No problem!

>
> Now, as for the ABI discussion above, I decided to try
> abi-compliance-checker on 2.8.11 vs branch 2.8 and 2.8.0 vs branch
> 2.8. I had to patch 2.8.0 with
> http://trac.wxwidgets.org/changeset/61009 or it wouldn't compile, but
> that's a trivial change.
>
> I compiled with --enable-unicode and --enable-debug, and I'm happy
> to report that both the headers and shared objects tests were
> successful, both against 2.8.0 and 2.8.11.
>
> I tried with --disable-debug, and it reported success as well, but
> showed some warnings about problems compiling the headers (for all
> versions). The .so files compared fine.

That's excellent, thank you very much - that gives us a bit more
confidence. Cool tool...

> Let me know if you want me to test any other linux configurations.

I would think that's probably enough, thanks.

>
> abi-compilance-checker generates HTML reports, I'll be happy to
> attach them or upload them to my server if you want to see them, but
> they basically list the headers and libraries and the compatibility
> problems, and those are all 0 for all tests.

Great! No need to upload the reports as far as I'm concerned.

Best regards,

Julian
>
> Regards,

[Vadim Zeitlin]

On Sat, 19 Mar 2011 16:10:18 -0500 Angel Vidal <cafe...@gmail.com> wrote:

AV> I tried with --disable-debug, and it reported success as well, but showed
AV> some warnings about problems compiling the headers (for all versions). The
AV> .so files compared fine.

Thanks a lot for doing these tests! I don't know if it's worth uploading
all the reports but I'd be interested to know what were these warnings about?

Thanks,
VZ

[chris elliott]

Hi, 2.8.12 is now online at http://biolpc22.york.ac.uk/pub/2.8.12/

I#ll upload it to SF early next week when back in York.

chris

[Bryan Petty]

On Thu, Mar 24, 2011 at 10:33 AM, chris elliott <bio...@york.ac.uk> wrote:
> Hi, 2.8.12 is now online at http://biolpc22.york.ac.uk/pub/2.8.12/
>
> I#ll upload it to SF early next week when back in York.

Thanks Chris, I appreciate your time on this.

I'll push the new docs up online just as soon as it's on SF.

Regards,
Bryan Petty

[chris elliott]

yes, and i'll tag it too
chris