On 22/01/14 08:10, Anton Khirnov wrote: > > On Wed, 22 Jan 2014 00:42:10 +0100, Luca Barbato <[email protected]> wrote: >> On 21/01/14 19:28, Anton Khirnov wrote: >>> Libavdevice is not really a standalone library, it depends (by >>> definition) on libavformat internal symbols. As there are no real reason >>> for it to be a separate library, the best solution for the associated ABI >>> problems is to merge it with libavformat. >> >> Beside Derek articulated explanation on why it is a bad idea looks like >> that there isn't support for this. > > After a discussion on IRC: > Hendrik, J-B, wm4 and others don't mind this in principle, as long as > --disable-devices still works. > I failed to get any arguments from Kieran.
We'll ask him again. >> Personally I'd rather change the capture and playback api so it works >> correctly than merge it back to avformat. > > That would be completely unrelated to this patch though. Yep > The point of this patch is to eliminate some semi-public APIs. Unless you want > to either > a) rewrite the devices so they are not demuxers at all, but use a completely > different API (eliminating most of their usefulness IMO) > or > b) make a stable API and ABI for external demuxers > you would not solve the problem. Stable api for external components can be discussed for libav12 I guess. The only marginal problem is having loads of external libraries pulled in and that might or might not annoying certain people. lu _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
