On Thu, Nov 06, 2014 at 07:35:29PM +0000, Stuart Henderson wrote: > On 2014-11-06, Alexandre Ratchov <a...@caoua.org> wrote: > > the ossaudiodev python module doesn't seem used/useful. It exposes > > an os-dependent interface (we're about to drop), which makes python > > scripts using it as portable as C code using oss ioctl is. > > > > OK to remove it? > > it shows in a ports search in a few places. no objection in principle > to removing but we should check over these: > > audio/cplay - mixer (master, pcm volume)
this one uses external player (sox). The volume knob didn't work anyway, so no regression. No misbehaving caused by ossaudiodev removal. > games/renpy - mixer (pcm volume) sound works, couldn't find any volume knob though > net/gajim - play sound files, if an external player is not configured gajim seems to use sox by default (though i don't see it listed in it's RUN_DEPENDS), didn't notice any problems. > textproc/py-nltk - play sound files, fallback to pygame which uses sdl > x11/kde/utils3 (looks like an example program and probably not used) > x11/kde4/superkaramba (ditto, related to above) > > the module is still present in python 3 but we don't build it. > also the string appears in a few other ports for syntax highlighting > and docs, I didn't include those here. > Another options would be to keep ossaudiodev, but only the oss mixer ioctls(). As the plan is to handle mixer in a second pass, so we could remove it later. Still this may confuse programs (like py-ntlk) that would detect ossaudiodev presence and then try to use it instead of pygame or a working external program. Furthermore oss mixer ioctls() don't work (not even with azalia as first device), so I don't see any usefulness in keeping it. After all sooner or later python scripts will have to face the reality: oss is not portable, and openbsd's implementation is a ugly hack with no future.
