[Insight-developers] StimulateImageIOTest: Strange Test Failures

[Bradley Lowekamp]

That was a good find in the  
itkImageFileWriterStreamingPastingCompressions test. I am glad it was  
a real bug. Base on your finding of this bug. I wonder if I should  
also add the Binary/ASCII flags to the permutations tested too.

Brad


On Mar 22, 2009, at 7:04 PM, Bill Lorensen wrote:

> StimulateImageIO. I fixed the VTKImageIO the other day.
>
>
> On Sun, Mar 22, 2009 at 2:45 PM, Luis Ibanez  
> <luis.ibanez at kitware.com> wrote:
>> Bill,
>>
>> Is this referring to the StimulateImageIO class ?
>> or to the VTKImageIO class ?
>>
>> ...
>>
>> Thanks for tracking and fixing this.
>>
>>
>>    Luis
>>
>>
>> --------------------
>> Bill Lorensen wrote:
>>>
>>> Luis,
>>>
>>> I think I fixed it. The class was in a bit of a mess. If it could  
>>> not
>>> open a file for reading or writing, it just ignored the failed open
>>> silently. And it could not handle file paths with spaces. When it
>>> failed to open them, it still tried to read the file. This explains
>>> the failed tests on some platforms.
>>>
>>> I added exceptions for failed opens. Then the second test failed. It
>>> turned out, the BigEndian.spr file had ^M's in it. After I ran
>>> dos2unix on the file, the test passed.
>>>
>>> Now I think the tests should pass on all platforms. We'll see...
>>>
>>> Bill
>>>
>>> On Sun, Mar 22, 2009 at 12:50 PM, Bill Lorensen <bill.lorensen at gmail.com 
>>> >
>>> wrote:
>>>
>>>> Still a small problem. I'll find it shortly.
>>>>
>>>> On Sun, Mar 22, 2009 at 12:09 PM, Bill Lorensen <bill.lorensen at gmail.com 
>>>> >
>>>> wrote:
>>>>
>>>>> Luis,
>>>>>
>>>>> The print did it. It turns out that the code was using sscanf to  
>>>>> parse
>>>>> the data file name. If the data file name had an internal space,  
>>>>> the
>>>>> sscanf stopped there. I replaced the parsing code with a regular
>>>>> expression that removes leading and trailing spaces. I'm not  
>>>>> sure why
>>>>> the input code did not detect a bad filename.
>>>>>
>>>>> Checked in the changes. Also, yesterday I found the problem with  
>>>>> the
>>>>> VTK IO test that was failing. I think that leaves us with one  
>>>>> failing
>>>>> gdcm related test on SUN OS.
>>>>>
>>>>> Bill
>>>>>
>>>>> On Sat, Mar 21, 2009 at 7:09 AM, Luis Ibanez <luis.ibanez at kitware.com 
>>>>> >
>>>>> wrote:
>>>>>
>>>>>> More on this topic:
>>>>>>
>>>>>> * Running the test manually in dash19 (by using the  ctest -R  
>>>>>> command)
>>>>>> produces
>>>>>>  the same error that we see in the Dashboard
>>>>>>
>>>>>> * Running the test by typing the command and using two local  
>>>>>> files as
>>>>>> output
>>>>>> works fine. (no errors produced)
>>>>>>
>>>>>>
>>>>>> It looks like the problem is somehow related to the path where  
>>>>>> the
>>>>>> files are written.
>>>>>>
>>>>>> I added the filename to the Exception message that will be  
>>>>>> produced
>>>>>> from:
>>>>>>
>>>>>> File: C:\Dashboards\My Tests\Insight\Code\IO 
>>>>>> \itkStimulateImageIO.cxx
>>>>>> Line: 210
>>>>>>
>>>>>> We will see a bit more about this message in tomorrow's  
>>>>>> dashboard.
>>>>>>
>>>>>>
>>>>>>   Luis
>>>>>>
>>>>>>
>>>>>> ------------------------------------------------
>>>>>> On Sat, Mar 21, 2009 at 9:47 AM, Luis Ibanez <luis.ibanez at kitware.com 
>>>>>> >
>>>>>> wrote:
>>>>>>
>>>>>>> Hi Bill,
>>>>>>>
>>>>>>> Here is what I have found in dash19 so far:
>>>>>>>
>>>>>>> The test creates the pair:
>>>>>>>
>>>>>>> * test1.spr  (text header)
>>>>>>> * test2.sdt  (binary pixel data)
>>>>>>>
>>>>>>> when copying these two files to my Linux machine,
>>>>>>> and comparing them with the ones produced by the
>>>>>>> Linux build, it happens that:
>>>>>>>
>>>>>>> * The two binary files are identical
>>>>>>> * The header files only differ in the filename of
>>>>>>>  the binary file. (which is normal, since the text
>>>>>>>  header stores a full path to the binary file.
>>>>>>>
>>>>>>> The binary build in dash19 has a space in the path,
>>>>>>> but... that's also the case for most of other builds..
>>>>>>>
>>>>>>> I'll try now running the test manually in dash19...
>>>>>>>
>>>>>>>
>>>>>>>   Luis
>>>>>>>
>>>>>>>
>>>>>>> -------------------------------------------------------------
>>>>>>> On Wed, Mar 18, 2009 at 3:04 PM, Bill Lorensen
>>>>>>> <bill.lorensen at gmail.com> wrote:
>>>>>>>
>>>>>>>> Luis,
>>>>>>>>
>>>>>>>> itkStimulateImageIOTest fails on two different compilers on  
>>>>>>>> the same
>>>>>>>> machine: dash19.kitware.
>>>>>>>>
>>>>>>>> http://www.cdash.org/CDash/testDetails.php?test=18814317&build=294262
>>>>>>>> http://www.cdash.org/CDash/testDetails.php?test=18799150&build=294031
>>>>>>>>
>>>>>>>> Both tests fail in the same way. The file they fail to read is
>>>>>>>> created
>>>>>>>> by the test.
>>>>>>>>
>>>>>>>> Perhaps you can run them manually? Try removing all the files  
>>>>>>>> in the
>>>>>>>> temp directory first.
>>>>>>>>
>>>>>>>> Bill
>>>>>>>>
>>>>>>>
>>>
>>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.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