OK. Sorry all. The 5.7 release of OpenBSD came out Friday evening so I had to get some upgrades done. I'm going to dive back into this and see how far I get.
Thanks, Bryan On Wed, Apr 29, 2015 at 1:43 PM, Kurt Miller <k...@intricatesoftware.com> wrote: > On Tue, 2015-04-28 at 20:47 -0400, Bryan C. Everly wrote: > > Kurt, > > > > > > I'm wrapping my head around gnu autoconfigure tonight. > > > > > > It looks like we need to modify the .m4 files in the common/autoconf > > directory and then it generates the generated-configure.sh script from > > there, right? > > Yea, that sounds right. > > > If that's the case, I can start working on patches to the *.m4 files > > to seed the diffs you sent from generated-configure.sh if you are OK > > with that (I know we need to wait until my userid shows up on the > > contributor list before we can do any of this officially). > > Sure we can share diffs here now. When your name shows up on the list, I > can start committing the ones that are ready. > > > Next question - how much can we mess with the default "bsd" stuff? > > Should we be doing specific things in an "openbsd" block so as not to > > mess up the FreeBSD port (my assumption is that this mercurial tree > > builds successfully under FreeBSD - right)? > > Right. For the areas where OpenBSD differs, we need to isolate those > changes from FreeBSD, NetBSD, Linux, Solaris, etc. While our trees are > separate from the main tree, we aspire to keep our changes in a state > that would be acceptable and non-breaking to other platforms so they > could be merged back into the main trees. > > Note that many times the generic bsd category blocks are sufficient and > there's no need to call out duplicate OpenBSD blocks when the bsd one > works. > > > I'm thinking that's how we should be operating. If you are ok with > > that, i'll start making the necessary changes to get the common things > > in "bsd" and the things that are unique to OpenBSD in their own > > sections. > > Right. We're on the same page. > > > I think I can get the freetype stuff fixed (if I'm following the *.m4 > > files correctly) as well as the location for giflib straightened out. > > Okay. I was thinking that perhaps a try_link test would be better for > freetype libs then the current filename comparison and it likely could > be the same for all bsd. > > > Finally, the myriad of --with* settings on the command line I'm > > guessing need to be baked into the *.m4 scripts if we are running on > > in OpenBSD? If so, then I can tackle getting that to work right. > > I'm not sure. Look to Linux and macos to see if those settings are > defaulted someplace. I would mimic their behavior here. Meaning if > building on linux or macos, the large set of --with settings are needed > there too, then I think we should conform to that pattern here too. > > In the end an OpenBSD port makefile will set all the env vars, configure > args, etc and its not really an issue to have a large set of configure > args in general. > > > What do you think? Am I on the right track? > > Yes for sure. > > Regards, > -Kurt > > >