[Insight-developers] Failing Tests after checkins

[Luis Ibanez]

Hi Dan,

No problem, I'll look at reducing the image size.

I agree that 500K is not that big in the medical image context,
but when we consider that this is testing infrastructure for
software, then it is hard to justify the ratio of code/images.


Maybe we should consider following the example of VTK and
having a separate ITKData checkout with all the additional
data for the testing infrastructure.


In this way we don't have to compromise the first user
experience versus the completeness of the testing suite.


   Regards,


      Luis


------------------
Dan Mueller wrote:
> Hi Luis,
> 
> Sorry for the "large" images. I guess I'm too used to working with
> 1GB+ images, so <500K is a drop in the ocean.
> 
> I tend to do my ITK stuff on the weekend; work keeps me busy enough
> during the week. I don't think I'll get around to this until then,
> which sounds too late. That means you will have to make the changes on
> your end. Sorry.
> 
> I don't like the idea of only testing the 2D code path, but I guess
> due to disk space we don't have an option... It would be nice to at
> least keep the 3D test for Hessian image function, seeing this will be
> what is most used.
> 
> Fire me an email if you have any questions when making the change.
> 
> Regards, Dan
> 
> 2008/10/28 Luis Ibanez <luis.ibanez at kitware.com>:
> 
>>Hi Dan,
>>
>>Thanks a lot for looking at the failing tests.
>>
>>It is great that you added the tests and baseline images too.
>>
>>However, the baseline images are larger than the typical
>>images that we use for regression testing.
>>
>>The images are:
>>
>>256K itkDiscreteHessianGaussianImageFunctionTest01_1.raw
>>256K itkDiscreteHessianGaussianImageFunctionTest01_0.raw
>>256K itkDiscreteGradientMagnitudeGaussianImageFunctionTest01.raw
>>256K itkDiscreteGaussianDerivativeImageFunctionTest01.raw
>>489K itkDiscreteHessianGaussianImageFunctionTest02_2.raw
>>489K itkDiscreteHessianGaussianImageFunctionTest02_1.raw
>>489K itkDiscreteHessianGaussianImageFunctionTest02_0.raw
>>489K itkDiscreteGradientMagnitudeGaussianImageFunctionTest02.raw
>>489K itkDiscreteGaussianDerivativeImageFunctionTest02.raw
>>
>>They account for about 3.5 Mb, which is too heavy to be
>>carried along with source code.
>>
>>To give you a reference, the full checkout of ITK is 139Mb...
>>
>>
>>Could we modify the regression test to run on 2D images
>>(instead of 3D ones), and then use only one of the slices
>>of these datasets ?
>>
>>
>>We would like to do this before cutting ITK 3.10 on Friday.
>>If your schedule is charged this week, just let us know,
>>and we will be glad to make the changes here.
>>
>>
>>
>>Please let us know,
>>
>>
>> Thanks
>>
>>
>>    Luis
>>
>>
>>
>>--------------------
>>Dan Mueller wrote:
>>
>>>Hi Bill, Luis,
>>>
>>>Sorry for the failures. It seems Luis beat me to the style issues. I
>>>have now fixed the tests. In the future I make sure to submit a
>>>successful experimental build before committing.
>>>
>>>
>>>
>>>>You should be receiving automatic e-mails regarding these issues.
>>>
>>>
>>>I am not receiving automatic e-mails. Is there somewhere I need to
>>>subscribe for these?
>>>
>>>Regards, Dan
>>>
>>>2008/10/25 Bill Lorensen <bill.lorensen at gmail.com>:
>>>
>>>
>>>>Dan,
>>>>
>>>>Did the tests pass on your machine before your checkins to Review?
>>>>http://www.cdash.org/CDash/viewTest.php?onlyfailed&buildid=204399
>>>>
>>>>Also, don't forget the warnings:
>>>>http://www.cdash.org/CDash/viewBuildError.php?type=1&buildid=204318
>>>>
>>>>And the Style violations:
>>>>http://www.cdash.org/CDash/testDetails.php?test=11452404&build=204318
>>>>
>>>>You should be receiving automatic e-mails regarding these issues.
>>>>
>>>>Regards,
>>>>
>>>>Bill
>>>
>>>
>