[Insight-developers] Dashboard Cleanup: mini0.nlm MacOSX-cross-rosetta : 50 tests failing

[Luis Ibanez]

Hi Michel,

Could you please take a look at the following build warnings:

/.../Insight/Utilities/NrrdIO/reorder.c:248: warning: array subscript
is below array bounds


They are reported in several Linux  machines (about 10 of them)
in today's dashboard.


It looks like a real bug.


    Thanks


         Luis


-------------------------------------------------------
On Wed, Jan 27, 2010 at 6:15 PM, Michel Audette
<michel.audette at kitware.com> wrote:
> Hi gents,
>
> I'm about half-way through a re-commit of Sourceforge Teem files, with
> the exception of sane.c and its enum in NrrdIO.h, which were isolated
> as the culprit (foreseeably, based on the errors that Brad was
> getting). See updated wiki:
> http://www.itk.org/Wiki/Proposals:Updating_Nrrd_library_2009
>
> I'm getting clean ctests so far on Linux and cross-compiled Mac:
> http://www.cdash.org/CDash/viewTest.php?onlypassed&buildid=524950
> http://www.cdash.org/CDash/viewTest.php?onlypassed&buildid=524973
>
> I'm also about to build and ctest the result on Windows 64 bits.
>
> My plan is to commit the rest of the files tomorrow, unless there is a
> problem on Windows, and revisit sane.c with you both after that.
>
> Best wishes,
>
> Michel
>
> On Wed, Jan 27, 2010 at 7:32 AM, Michel Audette
> <michel.audette at kitware.com> wrote:
>> Hi gents,
>>
>> it turned out that the new sane.c and its corresponding enum in
>> NrrdIO.h were the culprits. The latest commit of these changes caused
>> the problems on dashmacmini3 cross-compiled platform,
>> http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=524802
>> while other file commits up till then went fine.
>>
>> I will commit the old version of sane.c and NrrdIO.h, and propose to
>> commit other files, one by one, while validating with ctest as I'm
>> doing now with Linux and cross-compiled Mac builds, as well as
>> periodically on Windows.
>>
>> Gordon, perhaps you can take a fresh look at sane.c and the enum used
>> there, and arrive at a solution with Brad on the cross-compiled Mac?
>>
>> Cheers,
>>
>> Michel
>>
>> On Tue, Jan 26, 2010 at 4:22 PM, Bradley Lowekamp
>> <blowekamp at mail.nih.gov> wrote:
>>> I fired off the following:
>>> ctest -j2 -D Continuous -R Nrrd
>>> I am not entirely sure what that's going to do, but hopefully it'll give you
>>> the feedback you need.
>>> B
>>>
>>> On Jan 26, 2010, at 4:14 PM, Michel Audette wrote:
>>>
>>> Brilliant. I'll push on with more commits, and try to replicate what
>>> you were seeing, on the basis of these incremental commits.
>>>
>>> On Tue, Jan 26, 2010 at 4:07 PM, Bradley Lowekamp
>>> <blowekamp at mail.nih.gov> wrote:
>>>
>>> Hello,
>>>
>>> mini0 just finished. Looks good:
>>>
>>> http://www.cdash.org/CDash/buildSummary.php?buildid=524022
>>>
>>> Brad
>>>
>>> On Jan 26, 2010, at 2:55 PM, Michel Audette wrote:
>>>
>>> Any news, Brad?
>>>
>>> Cheers,
>>>
>>> Michel
>>>
>>> On Tue, Jan 26, 2010 at 1:27 PM, Bradley Lowekamp <blowekamp at mail.nih.gov>
>>>
>>> wrote:
>>>
>>> Working on the build now...
>>>
>>> On Jan 26, 2010, at 12:37 PM, Michel Audette wrote:
>>>
>>> Hi Brad,
>>>
>>> I've added a few NrrdIO files to yesterday's repository, and so far I'm
>>>
>>> not seeing NrrdIO-related ctest problems on dashmacmini3.
>>>
>>> Do you have time to do a check-out and give it a try? I'd like to proceed
>>>
>>> incrementally with your help as final arbiter.
>>>
>>> I'd like to commit a few more this afternoon, after I hear back from you.
>>>
>>> Cheers,
>>>
>>> Michel
>>>
>>> On Wed, Jan 20, 2010 at 12:09 PM, Bradley Lowekamp
>>>
>>> <blowekamp at mail.nih.gov> wrote:
>>>
>>> I quickly ran NrrdIO::airSanity(). It returned 7:
>>>
>>>  airInsane_AIR_NAN,       /*  7: airFPClass_f(AIR_QNAN,AIR_SNAN) wrong */
>>>
>>> So that mean that the endianness was correct.
>>>
>>> I am not sure what changes this update made that would cause this system
>>>
>>> to fail all NRRD IO.
>>>
>>> Brad
>>>
>>>
>>> On Jan 20, 2010, at 11:19 AM, Luis Ibanez wrote:
>>>
>>> Brad,
>>>
>>> Thanks a lot for looking into this.
>>>
>>> Is there something that we could do
>>>
>>> on our side to track the problem down ?
>>>
>>> Michel:   Could you please follow on this ?
>>>
>>>
>>>     Thanks
>>>
>>>
>>>            Luis
>>>
>>>
>>> -----------------------------------
>>>
>>> On Wed, Jan 20, 2010 at 9:43 AM, Bradley Lowekamp
>>>
>>> <blowekamp at mail.nih.gov> wrote:
>>>
>>> The occurred with the NRRD upgrade.
>>>
>>> Due to the oddness of the cross build,  (Building for ppc, but running
>>>
>>> on an intel) one of the first things I'd check is that the endianess is
>>>
>>> handled correctly. That is is should be a compile time check for apple
>>>
>>> systems. ITK does this correctly, but NRRD may do it differently.
>>>
>>> Brad
>>>
>>> On Jan 20, 2010, at 8:36 AM, Luis Ibanez wrote:
>>>
>>> Hi Brad,
>>>
>>> There are 50 tests failing in
>>>
>>>   mini0.nlm
>>>
>>>   MacOSX-cross-rosetta
>>>
>>>
>>> Most of them seem to be related to IO tests.
>>>
>>> http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=519324
>>>
>>>
>>> Do you have any hint on what may be the source of the problem ?
>>>
>>>
>>>   Thanks
>>>
>>>
>>>      Luis
>>>
>>>
>>>
>>>
>>>
>>>
>>> --
>>>
>>> Michel Audette, Ph.D.
>>>
>>> R & D Engineer,
>>>
>>> Kitware Inc.,
>>>
>>> Chapel Hill, N.C.
>>>
>>>
>>>
>>>
>>>
>>> --
>>>
>>> Michel Audette, Ph.D.
>>>
>>> R & D Engineer,
>>>
>>> Kitware Inc.,
>>>
>>> Chapel Hill, N.C.
>>>
>>>
>>> ========================================================
>>>
>>> Bradley Lowekamp
>>>
>>> Lockheed Martin Contractor for
>>>
>>> Office of High Performance Computing and Communications
>>>
>>> National Library of Medicine
>>>
>>> blowekamp at mail.nih.gov
>>>
>>>
>>>
>>>
>>>
>>> --
>>> Michel Audette, Ph.D.
>>> R & D Engineer,
>>> Kitware Inc.,
>>> Chapel Hill, N.C.
>>>
>>> ========================================================
>>>
>>> Bradley Lowekamp
>>>
>>> Lockheed Martin Contractor for
>>>
>>> Office of High Performance Computing and Communications
>>>
>>> National Library of Medicine
>>>
>>> blowekamp at mail.nih.gov
>>>
>>>
>>
>>
>>
>> --
>> Michel Audette, Ph.D.
>> R & D Engineer,
>> Kitware Inc.,
>> Chapel Hill, N.C.
>>
>
>
>
> --
> Michel Audette, Ph.D.
> R & D Engineer,
> Kitware Inc.,
> Chapel Hill, N.C.
>