[ITK] Win64 Visual Studio 12 segfaults

[Bradley Lowekamp]

Wow, very nice! Thanks for the clear commit message too.

Here is the latest dashboard build:
https://open.cdash.org/buildSummary.php?buildid=3650940

Way to squash 104 failing tests! One left though :)

With all this being done, I am still not sure I would trust this compiler to generate correct code with ITK.

Also has anyone tried the new VS13?

Brad

On Jan 12, 2015, at 11:31 AM, Matt McCormick <matt.mccormick at kitware.com> wrote:

> Excellent.  Thanks!
> 
> On Mon, Jan 12, 2015 at 11:26 AM, Floris Berendsen <floris at isi.uu.nl> wrote:
>> FYI: A workaround of Niels was submitted to gerrit (by Marius Staring).
>> Please have a look at the patch.
>> 
>> I8177d3b7: BUG: workaround for 64bit VS2013 Release segfault
>> http://review.source.kitware.com/#/c/18825/
>> 
>> Thanks,
>> Floris
>> 
>> On Fri, Jan 9, 2015 at 4:34 PM, Matt McCormick <matt.mccormick at kitware.com>
>> wrote:
>>> 
>>> Hi Niels,
>>> 
>>> Great news!
>>> 
>>> This same strategy was used to to fix some of the other failing tests
>>> [1]. While it should not strictly be necessary, it does hurt to
>>> improve our const-correctness and memory usage anyway.  Please
>>> consider submitting a patch for these changes [2].
>>> 
>>> Thanks,
>>> Matt
>>> 
>>> 
>>> [1]
>>> http://itk.org/gitweb?p=ITK.git;a=blobdiff;f=Modules/Filtering/ImageGrid/include/itkShrinkImageFilter.hxx;h=fca11a3eea3f885411192fd942c674cbba30c604;hp=86c3985601ff24ee36f6a0972be11573dfcc277c;hb=b92b952b98bee6dfc3837f4ed5d600e637cdcda9;hpb=7170ea77a918d268dd1971244ceb41b6d3518e21
>>> 
>>> [2]
>>> https://insightsoftwareconsortium.github.io/ITKBarCamp-doc/CommunitySoftwareProcess/SubmitAPatchToGerrit/index.html
>>> 
>>> On Fri, Jan 9, 2015 at 10:28 AM, Niels Dekker <niels-xtk at xs4all.nl> wrote:
>>>> Floris and I did some testing on how to work around this Visual C++
>>>> compiler bug. It looks like the crashes (segfaults) in both
>>>> FiniteDifferenceImageFilter::InitializeFunctionCoefficients() and
>>>> ImageRegion::IsInside(const Self&) disappear when a local object is
>>>> replaced by a const-reference.
>>>> 
>>>> In "Core\FiniteDifference\include\itkFiniteDifferenceImageFilter.hxx"
>>>> (line 273):
>>>> 
>>>> - const SpacingType spacing = outputImage->GetSpacing();
>>>> + const SpacingType & spacing = outputImage->GetSpacing();
>>>> 
>>>> 
>>>> In "Core\Common\include\itkImageRegion.h" (line 260):
>>>> 
>>>> - SizeType&  size = region.GetSize();
>>>> + const SizeType & size = region.GetSize();
>>>> 
>>>> 
>>>> I must admit I don't know why such a modification actually does suppress
>>>> the crashes. But it seems to do the job! What do you think?
>>>> 
>>>> Kind regards, Niels
>>>> 
>> 
>> 
> _______________________________________________
> Community mailing list
> Community at itk.org
> http://public.kitware.com/mailman/listinfo/community