[Insight-developers] Testing data addition guidelines

[M Stauffer (V)]

Yes, I was asking about appropriateness of adding a number of baseline
images, rather than the methodology. Thanks Brad, we'll use baseline
images for our tests.

-M

>Xiaoxiao, I think Michael was asking about whether use of 
>reference images
>is appropriate for his tests, not about how to actually add 
>the image data.
>
>> On Mon, Sep 12, 2011 at 1:13 PM, M Stauffer (V) 
><mstauff at verizon.net <mailto:mstauff at verizon.net>> wrote:
>>
>>     With the new testing data framework, what are the 
>guidelines now for
>>     submitting test data?
>
>The new framework overcomes technical limitations that 
>previously prevented
>data from being added in certain cases (e.g. very large 
>files).  With those
>limitations out of the way we have not written any new 
>guidelines AFAIK.
>
> >     If I remember right, part of the goal of the new
>>     framework was to allow more testing data submission?
>
>Yes.
>
>>     We'll be putting
>>     together a number of tests that could use reference 
>images for comparing
>>     against test output, or they could use a hard-coded 
>image metric value
>>     to evaluate the output. The latter would use less 
>testing data memory,
>>     but sacrifice a central historical record of the 
>reference image itself.
>
>Many of our tests use manually verified "baseline" images.  
>This is encouraged.
>If you think they are of value to your test then do not 
>hesitate to add them.
>
>-Brad