[Insight-developers] ITK 3.10 : Release Preparation : Dashboard Status

Luis Ibanez luis.ibanez at kitware.com
Tue Nov 4 12:24:20 EST 2008 

Yeap, it seems that having the implementation in a .cxx is what
will compartmentalize the instantiation of the map.

We will have to use a non-templated base class for the ResourceProbes.
It starts looking like too much trouble for simply using an
STL container   ...  :-/

I'll proceed to tag, and we will look at this after the
release.


    Thanks


       Luis


------------------------
Bill Lorensen wrote:
> Luis,
> 
> I believe the internal class must be in a .cxx so that it will be in
> the library. I suggest we address this after the tag. I'll have access
> to my VS6 build on Thursday.
> 
> Bill
> 
> On Mon, Nov 3, 2008 at 10:29 PM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
> 
>>Bill,
>>
>>Giving it a try to the method used in the MetaDataDictionary,
>>an internal class deriving from the std::map has been added
>>to the ResourceProbesCollectorBase class.
>>
>>Also, both the trait declaration and the m_Probes member variable
>>have been moved from the public & protected sections to the private
>>section, where BTW they should have always been.
>>
>>http://www.itk.org/cgi-bin/viewcvs.cgi/Code/Common/itkResourceProbesCollectorBase.h?root=Insight&r1=1.1&r2=1.2&sortby=date
>>http://www.itk.org/cgi-bin/viewcvs.cgi/Code/Common/itkResourceProbesCollectorBase.txx?root=Insight&r1=1.1&r2=1.2&sortby=date
>>
>>This second move may help to isolate any potential future API
>>changes related to the map.
>>
>>We should see the effect of this changes in the continuous build
>>from dash13 tomorrow morning.
>>
>>If it doesn't make a difference, we can always tag today's
>>CVS checkout.
>>
>>
>>
>>   Luis
>>
>>
>>----------------------
>>Bill Lorensen wrote:
>>
>>>My only concern is that the API of ResourceProbesCollectorBase may
>>>change to fix this problem. I think the same approach taken in
>>>MetaDataDictionary is needed. I would help, but I won't be back home
>>>until Thursday.
>>>
>>>We can probably take the risk about changing the API in this case. It
>>>is probably unlikely someone will use it in the next couple of weeks.
>>>
>>>I agree we should tag.
>>>
>>>Bill
>>>
>>>On Mon, Nov 3, 2008 at 2:17 PM, Luis Ibanez <luis.ibanez at kitware.com>
>>>wrote:
>>>
>>>
>>>>Hi Bill,
>>>>
>>>>I'm trying to track the problem with the Deformable registration tests
>>>>in VS60.  The class that seems to be involved with a std::map in these
>>>>tests is the new class: ResourceProbesCollectorBase, where an std::map
>>>>is used for storing the ids of the different probes.
>>>>
>>>>Although, it is not clear why only the tests from
>>>>
>>>> DeformableRegistration12.cxx
>>>> DeformableRegistration13.cxx
>>>>
>>>>are failing.
>>>>
>>>>Since the same class is used in the following tests/examples as well:
>>>>
>>>> DeformableRegistration14.cxx
>>>> DeformableRegistration15.cxx
>>>> DeformableRegistration4.cxx
>>>> DeformableRegistration7.cxx
>>>> DeformableRegistration8.cxx
>>>>
>>>>
>>>>(...?)
>>>>
>>>>
>>>>It would take a bit more time to track and solve this issue...
>>>>
>>>>I would vote for tagging today, and logging a bug report about
>>>>this VS60 problem.
>>>>
>>>>
>>>>Any objections ?
>>>>
>>>>
>>>>Luis
>>>>
>>>>
>>>>
>>>>---------------------
>>>>Bill Lorensen wrote:
>>>>
>>>>
>>>>>Issue B: My cygwin build shows the same failure. The changes that
>>>>>occurred the day it started to fail were all style changes. I looked
>>>>>through them and did not see any obvious candidates for trouble.
>>>>>
>>>>>The Deformable and TimeProbes failure are related I believe. Something
>>>>>to do with the use of <map> in a DLL I think. We say a similar problem
>>>>>a few years ago in MetaDataDictionary. Here are the comment there:
>>>>>// Implement map's api. On some Micorsoft compilers, stl containers
>>>>>// cannot be exported. This causes problems when building DLL's.
>>>>>// Here we inherit privately from std::map and provide a simple
>>>>>// API. The implementation will be in the DLL.
>>>>>
>>>>>
>>>>>Ignore the bias field error. I see this pop up once in a while on
>>>>>various platforms.
>>>>>
>>>>>Bill
>>>>>
>>>>>On Sun, Nov 2, 2008 at 9:47 AM, Luis Ibanez <luis.ibanez at kitware.com>
>>>>>wrote:
>>>>>
>>>>>
>>>>>
>>>>>>       ITK 3.10 Release Status
>>>>>>       =======================
>>>>>>
>>>>>>
>>>>>>Here is a summary of the remaining Dashboard issues:
>>>>>>
>>>>>>
>>>>>>
>>>>>>1) gelas.megason.lab: TimeStampTest  failed with a Time out:
>>>>>>http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=207544
>>>>>>May it look like a threading-related problem...?
>>>>>>(this is the only machine where the test fails (today)).
>>>>>>
>>>>>>
>>>>>>http://www.cdash.org/CDash/testSummary.php?project=2&name=itkTimeStampTest&date=2008-11-02
>>>>>>and the tests was passing before in this machine:
>>>>>>http://www.cdash.org/CDash/testDetails.php?test=11800122&build=207544
>>>>>>This is a 64bits machine.
>>>>>>
>>>>>>
>>>>>>2) dash21.kitware : Disk Full
>>>>>>Not really an ITK problem.
>>>>>>
>>>>>>
>>>>>>3) crl.med.hardvard.edu 64bits ICC 10  Debug
>>>>>>http://www.cdash.org/CDash/viewBuildError.php?type=1&buildid=207498
>>>>>>This seems to be a valid warning.
>>>>>>A virtual function shadowing another one in the base class.
>>>>>>Any objections about fixing this one ?
>>>>>>
>>>>>>
>>>>>>4) crl.med.hardvard.edu 64bits ICC 10  Release
>>>>>>http://www.cdash.org/CDash/viewBuildError.php?type=1&buildid=207534
>>>>>>same as (3)
>>>>>>
>>>>>>
>>>>>>5) RogueResearch3 : Mac10.5-InsightReview-gcc-dbg-ppc
>>>>>>http://www.cdash.org/CDash/viewBuildError.php?type=1&buildid=207495
>>>>>>Same as (3,4)
>>>>>>
>>>>>>
>>>>>>6) Dash14.kitware : Win32-cygwin344
>>>>>>http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=207702
>>>>>>ResampleImageFilter9Test : Failed
>>>>>>This is the only machine where the test is failing today:
>>>>>>
>>>>>>
>>>>>>http://www.cdash.org/CDash/testSummary.php?project=2&name=ResampleImageFilter9Test&date=2008-11-02
>>>>>>The test started failing in this machine between Oct 17 and 18th:
>>>>>>http://www.cdash.org/CDash/testDetails.php?test=11823849&build=207702
>>>>>>
>>>>>>
>>>>>>
>>>>>>7) dash13.kitware:  Win32-vs60
>>>>>>DeformableRegistration tests are failing (throwing Exceptions):
>>>>>>
>>>>>>DeformableRegistration12Test1
>>>>>>DeformableRegistration12Test2
>>>>>>DeformableRegistration12Test3
>>>>>>DeformableRegistration12Test4
>>>>>>DeformableRegistration13Test1
>>>>>>DeformableRegistration13Test2
>>>>>>DeformableRegistration13Test3
>>>>>>DeformableRegistration13Test4
>>>>>>
>>>>>>This is the only machine where these tests are failing today:
>>>>>>
>>>>>>
>>>>>>http://www.cdash.org/CDash/testSummary.php?project=2&name=DeformableRegistration12Test1&date=2008-11-02
>>>>>>
>>>>>>and the
>>>>>>itkTimeProbesTest (also throwing an Exception):
>>>>>>http://www.cdash.org/CDash/testDetails.php?test=11523633&build=207694
>>>>>>
>>>>>>
>>>>>>8) BillsLaptop: Win32-bcc5.5-Static
>>>>>>MRIBiasFieldCorrectionFilterTest:  Failed
>>>>>>This is the only machine where the test fails:
>>>>>>
>>>>>>
>>>>>>http://www.cdash.org/CDash/testSummary.php?project=2&name=itkMRIBiasFieldCorrectionFilterTest&date=2008-11-02
>>>>>>It seems to be a numerical issue:
>>>>>>"Absolute Avg. error with input and output mask = 33.6037"
>>>>>>"ERROR: sumOfError: 68171 is less than sumOfErrorFinal: 453650"
>>>>>>
>>>>>>In redwall.kitware (for example) this test passes with
>>>>>>the following equivalent output:
>>>>>>"Absolute Avg. error with input and output mask = 0.973767"
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=207427
>>>>>>
>>>>>>
>>>>>>----
>>>>>>
>>>>>>Suggested action items:
>>>>>>
>>>>>>From the list above, the problems that seem to be worth addressing
>>>>>>before the relase are:
>>>>>>
>>>>>>
>>>>>>A) itkImageToVectorImageFilter::SetNthInput() method issue.
>>>>>>
>>>>>>B) ResampleImageFilter9Test failure in cygwin.
>>>>>>(we need another cygwin build to compare against..)
>>>>>>
>>>>>>
>>>>>>
>>>>>>----
>>>>>>
>>>>>>
>>>>>>Comments and suggestions are welcome
>>>>>>
>>>>>>
>>>>>> Luis
>>>>>>
>>>>>>_______________________________________________
>>>>>>Insight-developers mailing list
>>>>>>Insight-developers at itk.org
>>>>>>http://www.itk.org/mailman/listinfo/insight-developers
>>>>>>
>>>>>
>>>>>
>

More information about the Insight-developers mailing list