llvm dev - Jul 2012 - [LLVMdev] Recent regression in CLANG/LLVM

Logged this: http://llvm.org/bugs/show_bug.cgi?id=13353

I might respectfully suggest that building ITK (like building Boost)
is an excellent torture test for CLANG.  Up until last week the SVN
head worked fine, now it doesn't.  Later today I'll try building ITK
again with CLang SVN head and see if someone fixed this...

Hi Kent, maybe it could be added to LLVM's nightly testsuite?

Ciao, Duncan.

On 17/07/12 17:57, Kent Williams wrote:
> Logged this: http://llvm.org/bugs/show_bug.cgi?id=13353
>
> I might respectfully suggest that building ITK (like building Boost)
> is an excellent torture test for CLANG.  Up until last week the SVN
> head worked fine, now it doesn't.  Later today I'll try building
ITK
> again with CLang SVN head and see if someone fixed this...
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

I'm not familiar with LLVM/CLANG's testing infrastructure, and don't
have the time myself to experiment with it.

ITK has an extensive regression test suite, based around CMake.  The
ITK consortium actually has a rule for new additions to ITK -- they
can't be integrated with ITK unless they're accompanied by regression
tests. They aren't always very sophisticated tests, but the code
coverage of the regression tests is over 80%

We have systems that do nightly build+test runs using the CLang that
comes with XCode on OS X.

CTest and the CDash web testing dashboard.  I could very easily set up
a CTest dashboard for CLang SVN Head compiling ITK and running ITK's
dashboard, but I don't know how that could be integrated with whatever
testing framework is in place for CLANG/LLVM.



On Tue, Jul 17, 2012 at 11:20 AM, Duncan Sands <baldrick at free.fr>
wrote:
> Hi Kent, maybe it could be added to LLVM's nightly testsuite?
>
> Ciao, Duncan.
>
> On 17/07/12 17:57, Kent Williams wrote:
>> Logged this: http://llvm.org/bugs/show_bug.cgi?id=13353
>>
>> I might respectfully suggest that building ITK (like building Boost)
>> is an excellent torture test for CLANG.  Up until last week the SVN
>> head worked fine, now it doesn't.  Later today I'll try
building ITK
>> again with CLang SVN head and see if someone fixed this...
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev

Update, Rafael Ávila de Espíndola just marked this fixed.  Cheers for
that. I think the assert fail dump that instructs you how to send in a
bug report is pretty awesome. The system works!

On Tue, Jul 17, 2012 at 10:57 AM, Kent Williams
<nkwmailinglists at gmail.com> wrote:
> Logged this: http://llvm.org/bugs/show_bug.cgi?id=13353
>
> I might respectfully suggest that building ITK (like building Boost)
> is an excellent torture test for CLANG.  Up until last week the SVN
> head worked fine, now it doesn't.  Later today I'll try building
ITK
> again with CLang SVN head and see if someone fixed this...