[CMake] FindX11.cmake needs some work for 2.6.0

[Bill Hoffman]

Alan W. Irwin wrote:
> On some platforms, X11_LIBRARIES derived by FindX11.cmake is going to be a
> mixture of full pathnames for libraries and a series of -l flags which is
> not a recommended form for CMake 2.6.x.  To see this, look at the
> FindX11.cmake logic for CMake-2.6.0-RC-8.
> 


OK, -l flags are fine with cmake 2.6.x, let me say that again, -l is OK 
with 2.6.  It MAY cause a link error.  If you set CMP0003 to NEW, it you 
can find out.  If CMP0003 is set to NEW and you stuff links, that is not 
a hack, that is the right way to do it.  We just did not want to break 
projects by default.  We wanted to warn by default, and give a way to 
silence the warning.  You do not have to use a full path.

> The preliminary version of X11_LIBRARIES is fullpath only since that is
> derived using FIND_LIBRARY, but then more libraries are added to this list
> as follows:
> 
> SET(X11_LIBRARIES ${X11_X_PRE_LIBS} ${X11_LIBRARIES} ${X11_X_EXTRA_LIBS})
> 
> X11_X_PRE_LIBS is also fullpath since it is the result of FIND_LIBRARY, but
> X11_X_EXTRA_LIBS (empty on my Debian testing system) potentially has 
> lots of
> -l flags such as -lnsl_s -linet -ldnet, etc. if you follow the
> X11_X_EXTRA_LIBS logic in FindX11.cmake.
> 
> Could somebody please change FindX11.cmake so that X11_X_EXTRA_LIBS is
> always determined with FIND_LIBRARY?  This would make it and X11_LIBRARIES
> consistently fullpath on all platforms.

I don't think I am going to make a change for 2.6 like this as it is 
likely not to work every where it does now.   Also, it is not required 
to have full paths.

> 
> I think these -l forms of libraries just slipped through as some ancient
> (autoconf?) boilerplate for X11_X_EXTRA_LIBS that has propagated to all
> FindX11.cmake versions in contradiction to the recommendation that 
> libraries
> are supposed to be in fullpath form for 2.6.x.
> 
> Furthermore, making X11_LIBRARIES fullpath would benefit the PLplot build
> (and presumably other builds as well). When first setting up the 
> CMake-based
> build system for PLplot, I either copied from our previous autotools-based
> build system or I may have noticed the potential for -l flags in
> X11_LIBRARIES.  But in any case, I prepended X11_LIBRARIES with
> "-L${X11_LIBRARY_DIR}" so that our users could have a non-standard install
> location for X (especially important for this era of frantic X development
> activity where some X testers may have multiple versions of X installed on
> their platform). However, this -L flag leads to CMP0003 warnings for the
> 2.6.0 release candidate since fullpath libraries are preferred for 2.6.x.
> (Also, X11_LIBRARY_DIR is an undocumented variable for FindX11.cmake which
> I would prefer not to depend upon in the long term.)
> 
> Of course, there are workaround for CMP0003 warnings, but the clean 
> solution
> would be if a CMake developer would simply change FindX11.cmake in cmake 
> cvs
> and also for 2.6.0 so that X11_X_EXTRA_LIBS and therefore X11_LIBRARIES are
> always consistently fullpath.  That would allow me to stop prepending
> "-L${X11_LIBRARY_DIR}" to X11_LIBRARIES which would eliminate the CMP0003

You should never use -L directly.....  link_directories should be used.

> warnings without compromising the need of some who require a non-standard
> install location for X.
> 
I guess I would suggest that you should experiment with a new version of 
FindX11 in plplot.  If you find it works well, we can look at it for 
inclusion in 2.6.1.  In the long run full paths would be a good idea for 
FindX11, but not a requirement right now.

Thanks.


-Bill