Hi Omar! On Tue, 2021-11-16 at 11:36 +0100, Omar Polo wrote: > there's an extraneous file `build' in the tarball, forgot to delete, > please ignore it :) > > Omar Polo <[email protected]> writes: > > > Clemens Gößnitzer <[email protected]> writes: > > > > > Attached is a new port, math/metis. > > > > > > portcheck complains about the hardcoded /usr/local/bin/gmake. However, if > > > I set > > > it to ${MAKE_PROGRAM}, configure fails with "env: ./gmake: No such file or > > > directory". How can I improve this? > > > > > > Additionally, make lib-depends-check still complains and errors out, > > > although I > > > added all the necessary libraries to WANTLIB. > > > > > > Note: Together with math/scotch that I already sent out, this enables me > > > to > > > successfully compile OpenFOAM (www.OpenFOAM.org), which I will port next. > > > > > > Thanks. > > > > No offense intended, but upstream is a bit crazy :) > > > > They use gmake to drive the cmake build. Fortunately, we can avoid > > that.
Thanks for this improvement. > > > > Here's an improved tarball which uses MODULES = devel/cmake and passes > > the correct flags via CONFIGURE_ARGS so cmake can build. As a bonus > > point, half of the diff to libmetis/CMakeList.txt isn't needed anymore, > > as modcmake take cares of the library verisioning. > > > > While here, I've also added NO_TEST=Yes since there aren't tests. > > > > Cheers, > > > > Omar Polo > > Updated patch attached, with no extra files and some whitespace cleanup. Builds and packages fine for me. Thanks.
metis.tgz
Description: application/compressed-tar
