On Wed, Apr 9, 2014 at 10:21 PM, Thomas Moulard wrote: > On Thu, Apr 10, 2014 at 1:48 PM, William Woodall > wrote: > > I would say no, I think it should set the flag in the rosdebian case and > do > > nothing in the other case, because if we have to set the value for the > plain > > debian case then why do we have GNUInstallDirs.cmake being used at all? > > Thanks for the feeback. See here then (again, untested for now): > > https://github.com/thomas-moulard/bloom/commit/00e977421dc86bfffc09ec51a2ae19d49a80212d Other than implementation details which I commented on, I think that would be fine. > > > Would this one be ok? > > Regarding GNUInstallDirs.cmake, the goal of this script is to make > CMake install step > more configurable. The part where it tries to guess libdir (and fail) > is for me a big mistake in the > design of this macro. For instance, there is no way I want multiarch > support for the software > I install in my home, even if I am running Debian :) > > > Also, this assumes that when you are running bloom you are on a debian > > machine. > > Ok I didn't know cross-releasing was allowed/supported by Bloom. > > [1] > https://www.debian.org/doc/debian-policy/ch-controlfields.html#s-f-Standards-Version > -- > Thomas > -- William Woodall ROS Development Team william@osrfoundation.org http://williamjwoodall.com/