I would suggest trying to get upstream to accept any changes you made to
the build system to get it to build on OpenBSD, but as for adding it to
ports, this is pretty insane. Between their dependency handling and lack of
release tarballs ("fetch the code from git", yawn) their infrastructure
is really hostile to 3rd-party packagers.We have versions of a lot of the dependencies *with patches to make them work properly on OpenBSD*, it would be really stupid to have to duplicate all this work. I could understand this for a handful of specialist libs, but really, they do this for everything: apr, apr-util, curl, ldns, libedit, pcre, portaudio, spandsp, speex, tiff, ... I do sort-of understand why they do this, but it's not at all helpful for packagers, and almost every other large project manages to do this reasonably well... On 2011-11-26, Ted Bullock <tbull...@comlore.com> wrote: > Hi Folks, > > So our company recently merged with a small telecom which has been > using pre-packaged versions of trixbox running on top of centos to > operate for the last while. > > However I am not happy with the reliability they have seen, as well > the security situation is rather poor. > > Anyway, I spent my afternoon banging my head against freeswitch's > build system making it build on OpenBSD 5.0 and I have produced some > working binaries. That said, the upstream installation process sucks, > and looking in seems really dumb (they have made a shwack of shallow > forks of most of their dependencies and pulled them all into their > tree). > > I don't really have any visibility on how deep some of those forks go, > but they are including really disparate stuff like sqlite and libtiff. > > Anyway, I'm willing to contribute paid developer time towards building > a freeswitch port and would be more than a little interested in what > else has been done towards this. Not much has appeared on marc.info > so I thought I'd ask. >
