[Ros-release] Dry stack prereleases for groovy

Tully Foote tfoote at willowgarage.com
Mon Nov 26 22:14:50 UTC 2012


So I think that with the change of gtest packaging to not distribute the
libraries we have updated our macros to build gtest on demand into every
project so that gtest gets the correct flags for the project.  (This was
the reason for not distributing gtest as a library.  )  It is my
understanding that gtest should be being automatically built locally and
automatically linked, not from /usr/ but from the build directory when you
use the gtest macros.   As such I think that all these workarounds are not
necessarily what we want, this should be handled by the macro itself.  Dirk
worked on this most recently.  He'll be back from vacation shortly.


On Mon, Nov 26, 2012 at 1:02 PM, Peter Soetens <peter at thesourceworks.com>wrote:

> On Mon, Nov 26, 2012 at 8:06 PM, Wim Meeussen <wim at hidof.com> wrote:
> > I had done the exact same thing and added the workaround to the Jenkins
> > setup preamble. This is the verbose output of what actually gets called:
> >
> >
> >
> >
> > /usr/bin/c++    -pthread -O2 -g
> > -L/tmp/install_dir/stack_overlay/unique_identifier/unique_id/lib
> > CMakeFiles/test_unique_id.dir/tests/test_unique_id.cpp.o  -o
> > ../bin/test_unique_id -rdynamic -L/opt/ros/groovy/lib -lroscpp
> > -lboost_signals-mt -lboost_filesystem-mt -lrosconsole
> -lroscpp_serialization
> > -lxmlrpcpp -lrostime -lboost_date_time-mt -lboost_system-mt
> > -lboost_thread-mt -lpthread -lcpp_common -l/usr/lib/libgtest.a
> > -Wl,-rpath,/opt/ros/groovy/lib
> >
> > /usr/bin/ld: cannot find -l/usr/lib/libgtest.a
>
> I think it would work with dropping the '-l' part, although my
> interpretation of the ld man page would say that
> -l:/usr/lib/libgtest.a would also work. So if you can hard-code that
> colon somewhere, you're done too.
>
> >
> >
> >
> >
> > So somewhere rosmake is using "-l/usr/lib/libgtest.a" instead of "-L
> > /usr/lib -lgtest". Any suggestions from a rosmake guru?
>
> You're of course waiting for Brian to take the bait :-) The latter
> form is considered bad practice. One should always use the absolute
> path and no search paths.
>
> Peter
> _______________________________________________
> Ros-release mailing list
> Ros-release at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-release
>



-- 
Tully Foote
tfoote at willowgarage.com
(650) 475-2827
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-release/attachments/20121126/e424bbeb/attachment-0009.html>


More information about the Ros-release mailing list