[Insight-developers] Problem with ITK_SOURCE_DIR var in ITKConfig.cmake.in

Stephen R. Aylward Stephen.Aylward at Kitware.com
Thu Oct 18 11:24:08 EDT 2007


That could work!  Great idea!

Also, just because it is undocumented, doesn't mean they won't use it. 
Otherwise much of ITK wouldn't have been used at the start :)   That is 
why we also have to be careful when making changes to stuff in Utilities 
- we cannot assume that such code is only used via ITK's access methods 
- some people could be using them from outside of ITK (e.g., some people 
may have projects that directly use Insight/Utilities/GDCM and nothing 
else from ITK - that is ok and we need to maintain backward 
compatibility for them too).

Thanks,
Stephen


Mathieu Malaterre wrote:
> Hi,
> 
>   I saw recently that a bunch of *LEGACY* macro were added to ITK.
> Maybe I can add ITK_SOURCE_DIR in between those *LEGACY*. That is to
> say ITK_SOURCE_DIR will only be accessible if LEGACY is ON.
> 
> whatdayasay ?
> 
> -Mathieu
> 
> On 10/18/07, Stephen R. Aylward <Stephen.Aylward at kitware.com> wrote:
>> However - you are right - we do need to fix bugs. :)
>>
>> (sometimes I come across sounding too strong, i.e., like a stubborn
>> idiot :) sorry)
>>
>> Perhaps there is a way to devise a backward compatible solution - or
>> throw a warning - or document the change?
>>
>> Thanks for your patience with my occasional rant,
>> Stephen
>>
>> Mathieu Malaterre wrote:
>>> Hi Stephen,
>>>
>>>   As far as I understand removing bugs do not need to be backward
>>> compatible. ITK_SOURCE_DIR was pointing to non-existant directory
>>> anyway. I don't even understand in which case it was returning
>>> something useful.
>>>
>>>   It was discussed on the dev mailing list right after 3.4 so this has
>>> never been part of any release. If you still feel that it should be
>>> back for 3.6, please let me know. But again this is undocumented
>>> 'feature' and high on the 'shoot yourself in the foot' list.
>>>
>>> Regards,
>>> -Mathieu
>>>
>>> On 10/18/07, Stephen R. Aylward <Stephen.Aylward at kitware.com> wrote:
>>>> Regardless, this broke backward compatibility.
>>>>
>>>> ITK's backward compatibility policy MUST also apply to its CMake files.
>>>>
>>>> Perhaps there was a really good reason for doing it, and it wasn't
>>>> removed just because it was "bad" - but regardless, this should have
>>>> been posted to the developers and users list, included in a CHANGES.txt
>>>> file, and announced in every viable way to keep these questions from
>>>> arising.
>>>>
>>>> Stephen
>>>>
>>>> Mathieu Malaterre wrote:
>>>>> On 10/18/07, Thomas Feuvrier <thomas.feuvrier at c-s.cnes.fr> wrote:
>>>>>> Hello,
>>>>>>
>>>>>> In the new version of ITK (3.4.0), we saw that the ITK_SOURCE_DIR
>>>>>> variable in ITKConfig.cmake.in has disapeared ! It's important for us,
>>>>>> becasue without this variable, my project (OTB) can't use the latest ITK
>>>>>> version. Is it an error ?
>>>>> Yes this is an error to use this variable. When importing an ITK
>>>>> build/install tree you should *not* rely on ITK_SOURCE_DIR since it
>>>>> would in general point to a non existant directory or worse to a
>>>>> different ITK than the one you are importing.
>>>>>
>>>>> HTH,
>>>> --
>>>> =============================================================
>>>> Stephen R. Aylward, Ph.D.
>>>> Chief Medical Scientist
>>>> Kitware, Inc. - Chapel Hill Office
>>>> http://www.kitware.com
>>>> Phone: (518)371-3971 x300
>>>>
>>>
>> --
>> =============================================================
>> Stephen R. Aylward, Ph.D.
>> Chief Medical Scientist
>> Kitware, Inc. - Chapel Hill Office
>> http://www.kitware.com
>> Phone: (518)371-3971 x300
>>
> 
> 

-- 
=============================================================
Stephen R. Aylward, Ph.D.
Chief Medical Scientist
Kitware, Inc. - Chapel Hill Office
http://www.kitware.com
Phone: (518)371-3971 x300


More information about the Insight-developers mailing list