[Insight-developers] Failing LabelMap on windows

[Bradley Lowekamp]

Hello,

I was unable to get the tests to pass with out the print statements. I tried manipulating the if statements in a variety of configurations, with out success. I am now wondering if there is some kind of stack corruption going on. But I am not finding much logic to what is occurring, and if anyone has a"theory" as to why the windows compiler is doing what its doing here I love to here it.

Brad

On Feb 23, 2011, at 12:14 PM, Lowekamp, Bradley (NIH/NLM/LHC) [C] wrote:

> Hello,
> 
> Apologies for the prior unprofessional e-mail. I got a little excited seeing these test pass!
> 
> So adding the following lines of code made all the tests pass.
> 
> diff --git a/Code/Review/itkAttributeUniqueLabelMapFilter.txx b/Code/Review/it
> index 0a5c7e8..81512f5 100644
> --- a/Code/Review/itkAttributeUniqueLabelMapFilter.txx
> +++ b/Code/Review/itkAttributeUniqueLabelMapFilter.txx
> @@ -134,6 +134,7 @@ AttributeUniqueLabelMapFilter<TImage, TAttributeAccessor>
>            {
>            if ( l.labelObject->GetLabel() > prev.labelObject->GetLabel() )
>              {
> +            std::cout << "reversing!" << std::endl;
>              keepCurrent = !m_ReverseOrdering;
>              }
>            else
> @@ -145,6 +146,7 @@ AttributeUniqueLabelMapFilter<TImage, TAttributeAccessor>
>            {
>            if ( attr > prevAttr )
>              {
> +                       std::cout << "reversing!" << std::endl;
>              keepCurrent = !m_ReverseOrdering;
>              }
>            else
> 
> 
> Yes, adding print statements made the test pass! So it looks like we have an optimizer/compiler bug. A more permanent solution should follow this afternoon!
> 
> 
> Brad :)
> 
> On Feb 23, 2011, at 12:01 PM, Lowekamp, Bradley (NIH/NLM/LHC) [C] wrote:
> 
>> WOOT WOOT!
>> 
>> On Feb 21, 2011, at 5:53 PM, Luis Ibanez wrote:
>> 
>>> Hi Brad,
>>> 
>>> The misbehavior in that Windows build
>>> happens in the file:
>>> 
>>> ITK/Code/Review/itkAttributeUniqueLabelMapFilter.txx
>>> 
>>> between lines 121 and 206.
>>> 
>>> something goes wrong in the selections of lines that
>>> are later introduced in "lines.push_back()" commands.
>>> 
>>> 
>>> We were able to better locate the error by using the
>>> attached image as input to the test.  This attached
>>> image is a cropped region of interest from the image
>>> that is currently used as input in the test.
>>> 
>>> 
>>> Essentially, it seems that a couple of lines that
>>> should have been eliminated, end up being reinserted...
>>> 
>>> 
>>> 
>>>     Happy bug hunting !
>>> 
>>> 
>>>          Luis
>>> 
>>> 
>>> ----------------------------------------------
>>> On Mon, Feb 21, 2011 at 12:09 PM, Bradley Lowekamp
>>> <blowekamp at mail.nih.gov> wrote:
>>>> Luis and Gaetan,
>>>> 
>>>> I heard that you had narrowed down the problem with the failing LabelMap tests on the windows systems:
>>>> http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=871555
>>>> 
>>>> Some one said that you had it narrowed down to about 100 lines of code. Can you guys share so I can look to see if I find anything?
>>>> 
>>>> Thanks,
>>>> Brad
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>> 
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>> 
>>>> Kitware offers ITK Training Courses, for more information visit:
>>>> http://kitware.com/products/protraining.html
>>>> 
>>>> Please keep messages on-topic and check the ITK FAQ at:
>>>> http://www.itk.org/Wiki/ITK_FAQ
>>>> 
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://www.itk.org/mailman/listinfo/insight-developers
>>>> 
>>> <SpotsLabeled1.png>
>> 
>> ========================================================
>> Bradley Lowekamp  
>> Lockheed Martin Contractor for
>> Office of High Performance Computing and Communications
>> National Library of Medicine 
>> blowekamp at mail.nih.gov
>> 
>> 
>> <ATT00001..txt>
> 
> ========================================================
> Bradley Lowekamp  
> Lockheed Martin Contractor for
> Office of High Performance Computing and Communications
> National Library of Medicine 
> blowekamp at mail.nih.gov
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20110223/9c4a5613/attachment-0001.htm>