On 2017/04/26 00:47, Jeremie Courreges-Anglas wrote: > "Ted Unangst" <t...@tedunangst.com> writes: > > > Jeremie Courreges-Anglas wrote: > >> > >> clang already recognizes (and ignores) -fforce-addr but errors out > >> on -fno-force-addr. This breaks the build of ports/databases/qdbm: > >> > >> > >> http://build-failures.rhaalovely.net/amd64-clang/2017-04-14/databases/qdbm.log > >> > >> Fixing the port is easy, but so is ignoring -fno-force-addr. > >> -fno-force-addr should probably be pushed upstream if the latter is > >> preferred. > >> > >> Thoughts? > > > > If the option is obsolete, I'd say we'd be doing the world a favor by > > upstreaming a patch to fix qdbm. > > I'm not sure what you mean by obsolete ("ignored"?). A bunch of > packages out there use -fforce-addr but since qdbm seems to be the only > user of -fno-force-addr, I guess there's not much point in adding > support for a no-op. (The port has been fixed since.)
10 years old, so I don't think it's upstream-able. What they're actually doing here is "-fforce-addr" by default and then adding "-fno-force-addr" on BSDs to disable it again.