[Insight-developers] Need to add images for new tests to Data -- how to do that in a Gerrit topic?

Marcus D. Hanwell marcus.hanwell at kitware.com
Thu Nov 4 09:49:53 EDT 2010


On Thu, Nov 4, 2010 at 8:49 AM, Wes Turner <wes.turner at kitware.com> wrote:
> Hey Marcus,
>
> On Wed, Nov 3, 2010 at 5:18 PM, Marcus D. Hanwell
> <marcus.hanwell at kitware.com> wrote:
>>
>> On Wed, Nov 3, 2010 at 4:55 PM, kent williams
>> <norman-k-williams at uiowa.edu> wrote:
>> > Subject line says it all.
>> >
>> > Have a couple of small images that I need to add to ITK/Data submodule
>> > --
>> > how do I add them so that the Gerrit topic I'm pushing can access them?
>> >
>> You don't need to, you could add them to the repository now and they
>> will not hurt anything. There is no ITK data repository in Gerrit, and
>> review of the binary image files wouldn't work. Can't people generate
>> these images themselves if they want to? There is some work at the
>> moment to make things easier here - Luis may comment more on that.
>>
>
> Are you suggesting that reviewers manually generate binary images specific
> to a test before performing a Gerrirt review of a new submission?  That's a
> pretty big bar.  My normal mode is to get a clean "Review" directory of ITK,
> download the patch, run a "ctest -D Experimental" to make sure all the tests
> pass, and while the experimental is running perform a code walk through. If
> I need to generate a new binary image before the tests will pass, the review
> time goes way up ... and any failure would require that I determine which of
> the image or the code are wrong ....
> If on the other hand, you are suggesting that the images be loaded into the
> data submodule and that the reviewers manually download the new images, then
> that would probably be workable, although finding a way to point this out to
> the reviewer while the patch is being retrieved would be a great help.

I was merely enumerating options - it seems reasonable to commit the
images to the data module before the review, and then push the data
module hash forward. I think the more ideal case is to use the new
MIDAS based approach, and specify the hash of the image in MIDAS that
the test should download, but that is not quite ready to go as far as
I know.

I am not a core ITK developer, and so I was hoping those who develop
ITK would chime in and I will help with the Git/software process side
where appropriate.

Marcus


More information about the Insight-developers mailing list