Possible regression with CMake 2.8.6 RC2

[Alexander Neundorf]

On Sunday, September 11, 2011 04:53:09 AM Michael Pyne wrote:
> Hi all,
> 
> Just writing to let you all know that a regression was reported by a Mageia
> user after CMake 2.8.6 rc2 was installed on his system.
> 
> kdelibs (4.7 branch, git) refused to link in kdecore/util based on this.
> The reason appears to be that -lpthread was not being used in the link
> command anymore. After some troubleshooting the user downgraded to 2.8.5
> and reported that his kdelibs build completed successfully.
> 
> The CMakeLists.txt for kdelibs/kdecore has the following for the affected
> component starting from line 389:
> 
> if(NOT WIN32)
>     set(kdecore_LIB_SRCS ${kdecore_LIB_SRCS} util/kshareddatacache.cpp)
>     set(kdecore_OPTIONAL_LIBS ${kdecore_OPTIONAL_LIBS} \
>                               ${CMAKE_THREAD_LIBS_INIT})
> else(NOT WIN32) ...
> 
> Raphael Kubo da Costa mentioned that the only relevant change in CMake
> appeared to be from commit fd61be71401ef9e0a241562fc31539273084deff [1].
> 
> KSharedDataCache definitely requires pthreads at this point so if our
> current CMake-fu for utilizing that is wrong just let me know the right
> way to do it, otherwise I think this is a 2.8.6 regression.
> 
> [1]
> http://www.cmake.org/gitweb?p=cmake.git;a=commit;h=fd61be71401ef9e0a241562f
> c31539273084deff

I just started checking my mails after coming back from vacation...

Has this been reported to the cmake list or the cmake bug tracker ?
If this problem is present in the latest cmake RC, please send a mail to the 
cmake mailing list.
Any such regression is in general seen as a bug by cmake, so please report it 
there.
One of the main rules in cmake development is that a cmake update must never 
break any existing builds.

Thanks
Alex