[cmake-developers] 185 - CMakeOnly.AllFindModules (Failed)

[David Cole]

On Mon, Feb 13, 2012 at 12:44 PM, Rolf Eike Beer <eike at sf-mail.de> wrote:
> Am Montag, 13. Februar 2012, 11:33:14 schrieb Bill Hoffman:
>> On  our linux release machine we are getting an error in the modules
>> that is preventing the release from being made.  The error is:
>>
>> CMake Error at CMakeLists.txt:58 (message):
>>    GNUPLOT_VERSION_STRING has unexpected content
>>
>> -- Configuring incomplete, errors occurred!
>> CMake Error at Test.cmake:11 (message):
>>    CMake failed to configure AllFindModules
>>
>> The gnuplot on that machine is sooo old it does not respond to --version...
>>
>> The linux is very old so that CMake is forward compatible with other
>> machines.
>
> I have no problem in removing this check (and in fact I will). This test is
> only that strict to find if there are actually versions out there where the
> version check does not work. Looks like there are gnuplot versions where this
> will never work, so the check will go.
>
> Eike
> --
>
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers


Eike,

I've got a test machine setup where I'm using the latest Xcode 4.3
from Apple, and I'm getting output like this from the AllFindModules
test:

  CMake Error at CMakeLists.txt:52 (message):
    GIT_VERSION_STRING has unexpected content 1.7.7.5 (Apple Git-26)

Is there a way to say "it's ok for the version string to have extra
stuff after the version number".... without breaking your concept of
"expected content" too much? Really, our expectation of a version
number string in CMake is anything that's useful with VERSION_LESS and
VERSION_GREATER comparisons. The above string is useful, and would be
valid for version comparisons.

Perhaps we could use the regex you're using now, but only care about
the *start* of the string?


Thanks,
David