[ITK] Win64 Visual Studio 12 segfaults

[Bradley Lowekamp]

Hi,

No just 26 of them that are deconvolution filters, not the fast marching/colliding fronts ones. Most of those others where fixed in today builds on next [1].

It's quite hard to maintain all those builds, and wrappers for 7 different languages, please pardon the red ness. The top master/package section should be green except for window 64 linking issue.

The BasicFilters.* failing tests are the C++ native code, and don't have the extra language to hide the abnormal program termination.

The error reported by gtest is:
unknown file: error: SEH exception with code 0xc0000005 thrown in the test body.

I truthfully haven't seen that error before, and googling it looks like its similar to a segfault, I believe I saw some of these filters failing before it ITK with this compiler.

I hope your patch addresses these issue too, but it could very well be one of the existing examples you demonstrated that is not covered by ITK tests.

Brad



[1] https://open.cdash.org/viewTest.php?onlyfailed&buildid=3673220

On Jan 29, 2015, at 6:05 PM, Niels Dekker <niels-xtk at xs4all.nl> wrote:

> Thanks for your reply, Bradley.
> 
> Are you sure that those 36 SimpleITK test failures are also caused by Win64 VS12 compiler bugs?  They don't seem to be caused by segfaults, whereas most of the 1296 test failures from 12 days ago were marked as SEGFAULT: https://open.cdash.org/viewTest.php?onlyfailed&buildid=3656652 (Win64-vs12-shared-next, 2015-01-17).
> 
> I see many test failures from other compilers and architectures as well, at https://open.cdash.org/index.php?project=SimpleITK&date=2015-01-29 Are these just SimpleITK issues?
> 
> If those 36 SimpleITK test failures are from another Win64 VS12 compiler bug, I would be very interested to see a minimal example. You know, the segfault Win64 VS12 compiler bug can be reproduced by a program of just 10 lines of code, without any #include.  (As I commented at http://connect.microsoft.com/VisualStudio/feedbackdetail/view/1060166).
> 
> Any way, early next week we'll hope to submit our ITK patch. Maybe that also fix some of the SimpleITK issues...  :-)
> 
>  Kind regards, Niels
> 
> Bradley Lowekamp wrote on Jan. 29, 2015:
>> I also have a SimpleITK build with Visual Studio 2013 (VS12)
>> for SimpleITK for Win64 build against the ITK release branch:
>> 2
>> https://open.cdash.org/buildSummary.php?buildid=3671881
>> 
>> It is down to 36 failures from 1287.  The CollidingFronts and
>> FastMarchingUpwindGradient are SimpleITK issues.
>> 
>> What's interesting is that there no report internal compiler error,
>> with 64-bit ids enabled. We a bunch of Deconvolution filters
>> failing put nothing else.
>> 
>> Well this builds confidence in correct code being generated, I
>> still would not give the generated code away to anyone and expect
>> it to work 100%.
>> 
>> I think we should likely add a CMake warning for this problematic
>> compiler and architecture.
> 
> _______________________________________________
> Community mailing list
> Community at itk.org
> http://public.kitware.com/mailman/listinfo/community

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/community/attachments/20150129/52b57d2e/attachment.html>