Rationale behind _cmake_skip_rpath choice in /etc/rpm/macros.cmake

Theodore Papadopoulo Theodore.Papadopoulo at sophia.inria.fr
Wed Oct 14 19:02:18 UTC 2009


I would like to understand why the file macros.cmake as distributed in 
fedora-10 defines:
%_cmake_skip_rpath -DCMAKE_SKIP_RPATH:BOOL=ON

I use cmake to build an rpm for a software that builds several libraries 
and binaries (based
on those libraries). In the spec file of my rpm I decided to add some
make test
into order to check at rpm build time that everything is OK (I agree 
that this is probably an overkill).
Disabling the rpath made all the checks to fail, so I added a:

%define _cmake_skip_rpath -DCMAKE_SKIP_RPATH:BOOL=OFF

at the beginning of my specfile.

But I wonder whether there is any risk in doing so.... Is there a risk 
of library interception if someone
re-creates a library in the ??

I suppose that the choice was made on purpose, I just want to assess the 
risk I'm taking in having
this approach. Otherwise, I'll just remove the test stuff from the spec 
file (as I found no other way
of doing the tests... Setting LD_LIBRARY_PATH before the make test does 
not seem to work
probably because ctest is careful in setting a clean environment for the 
tests....).

The only other solution I can think of would be putting the check in the 
%install section after having
done the make install  and testing with a chroot placed on 
$RPM_BUILD_ROOT (but I fear I need a
subshell for the rpm process to proceed after the test)  ?
Are there any advice on this ?

In advance, thank's a lot.




More information about the fedora-devel-list mailing list