On Thu, 2002-07-11 at 06:21, Colin Walters wrote: > [ This should really be on -mentors... ] > > On Wed, 2002-07-10 at 23:56, Bastien Nocera wrote: > > > His upstream is not quite up to scratch. The intltool programs aren't > > supposed to be in the tarball. They should be generated from their .in > > counterpart by configure. > > Ok. In that case, they should probably be deleted by the upstream > Makefile's 'clean' target, which his debian/rules should call. If No, if it's generated by configure, and necessary to build, it should be removed by a make distclean. Just like make clean shouldn't remove Makefiles. > upstream doesn't delete them in 'make clean', then Ian should just > remove them in the debian/rules clean target manually. Better use the distclean target. > That way they won't end up in the diff (which is generated by > dpkg-source *after* the 'clean' target is called, under a normal > invocation of dpkg-buildpackage). > > > As for config.guess and config.sub, I really don't know. > > They should be part of the upstream source, I promise :) $ tar tzf acme-1.99.5.tar.gz | grep "config" acme-1.99.5/acconfig.h acme-1.99.5/config.h.in acme-1.99.5/configure acme-1.99.5/configure.in I don't have any here... And the automake manual says (that's an autoconf macro provided by automake, if you wonder): AC_CANONICAL_HOST Automake will ensure that `config.guess' and `config.sub' exist. Also, the `Makefile' variables `host_alias' and `host_triplet' are introduced. See section `Getting the Canonical System Type' in The Autoconf Manual. So I think config.guess and config.sub are only used if configure needs to check the type of the system. It's not needed for metacity-setup, so there's no need to copy it. The copy in the template rules file is because a lot of upstreams were using ancient version of them and builds were failing on hppa, and some of these weird archs debian supports. Cheers -- /Bastien Nocera http://hadess.net
Attachment:
signature.asc
Description: This is a digitally signed message part