[Insight-developers] WrapITK must found ImageCompare to run the
tests
Luis Ibanez
luis.ibanez at kitware.com
Thu Sep 7 17:35:14 EDT 2006
Hi Gaetan,
The original itkTestMain.h is located in Insight/Code/Common.
I would suggest to put there also the itkTestDriver.cxx and
build a single library for it.
I assume that by having this function in a library you will be
able to use it from the target languages, is that correct ?
Please let us know,
Thanks
Luis
---------------------
Gaëtan Lehmann wrote:
>
> Luis,
>
> The itkTestDriver is mostly ready:
> http://voxel.jouy.inra.fr/darcsweb/darcsweb.cgi?r=itkTestDriver;a=headblob;f=/itkTestDriver.cxx
>
> I wouldn't be too surprised to read that there is still some small
> problems or some english mistakes, but it is ready to be tested.
>
> Can you tell me where I should put it in ITK ? It should be built even
> if c++ tests are not built, so the wrapper test can be run, and it can
> be use to implement other test, for the IJ for example
>
> Thanks,
>
> Gaetan
>
>
> Le Thu, 07 Sep 2006 14:44:50 +0200, Luis Ibanez
> <luis.ibanez at kitware.com> a écrit:
>
>>
>> Ok, I will add ImageCompare as a default binary to build in ITK,
>> then look at removing the duplicate functionality from the test
>> drivers.
>>
>> Hopefully we would be able to do this without having to touch the
>> CMakeLists.txt files in the Testing directories...
>>
>>
>> Luis
>>
>>
>> -----------------------
>> Gaetan Lehmann wrote:
>>
>>> On Thu, 07 Sep 2006 13:48:15 +0200, Luis Ibanez
>>> <luis.ibanez at kitware.com> wrote:
>>>
>>>>
>>>> Hi Gaetan,
>>>>
>>>> The C++ test drivers already have a built-in Image Compare
>>>> functionality. So, it seems that the problem is that it is not
>>>> available to the wrapping tests.
>>>>
>>>> It was from the C++ test drivers that we took the code for creating
>>>> ImageCompare when we put it in the suggested material for the Insigth
>>>> Journal.
>>>>
>>>> That being said, ImageCompare is such a basic tool, that I agree with
>>>> you that adding it to the default binaries to be built by ITK is a
>>>> good idea.
>>>>
>>>> Option (3) also sounds like a good option, and it seems that it offers
>>>> extra advantages. My suggestion would be to do (1) right now,
>>>> just because it is so easy (we just need to commit the file), and
>>>> then
>>>> proceed to also implement (3).
>>>>
>>>>
>>>> Would you like to undertake (3) ?
>>>>
>>> Ok. Can you add ImageCompare to ITK ?
>>> Where do you think ImageCompare and the new test driver should go in
>>> ITK ?
>>> Also, what do you think of moving the common code from ImageCompare
>>> and itkTestMain.h in a common file, to avoid duplicating code, and
>>> to make implementation of the new test driver easier ?
>>> Gaetan
>>>
>>>>
>>>> Luis
>>>>
>>>>
>>>> ----------------------
>>>> Gaetan Lehmann wrote:
>>>>
>>>>> Hi,
>>>>> To implement WrapITK's tests, I have used ImageCompare, which is
>>>>> not by default in ITK.
>>>>> There is several possible ways to go:
>>>>> (1) add ImageCompare to ITK
>>>>> (2) implement the image comparaison in the test drivers
>>>>> (3) implement a test driver in c++ usable for all the languages,
>>>>> and able to compare the images
>>>>> (1) is the simplest one, and would be much give ImageCompare by
>>>>> default to the user, so it would be easier to begin to prepare a
>>>>> contribution to the insight journal
>>>>> (2) will duplicate the image comparaison task - not so good
>>>>> (3) should be mainly an adaptation of the ITK test driver to set
>>>>> the environment needed to run the tests with the target
>>>>> languages. Also, I think that this test driver would be better
>>>>> for the insight journal than ImageCompare: we can run the test
>>>>> and validate its results in a single test; it is not possible
>>>>> with ImageCompare. Finally, it should be easy with this one to
>>>>> fix the lib path problem with java
>>>>> My preference goes to (3) - please let me know what you think
>>>>> about that
>>>>> Thanks,
>>>>> Gaetan
>>>>>
>>>>
>>>
>>
>>
>
>
>
More information about the Insight-developers
mailing list