On Mon, Mar 25, 2013 at 02:23:32PM +0200, Martin Storsjö wrote: > On Mon, 25 Mar 2013, Diego Biurrun wrote: > >On Mon, Mar 25, 2013 at 01:31:55PM +0200, Martin Storsjö wrote: > >>On Mon, 25 Mar 2013, Diego Biurrun wrote: > >>>On Mon, Mar 25, 2013 at 12:46:58PM +0200, Martin Storsjö wrote: > >>>>--- a/configure > >>>>+++ b/configure > >>>>@@ -1691,7 +1692,7 @@ wmv3_vdpau_decoder_select="vc1_vdpau_decoder" > >>>> > >>>># parsers > >>>>-h264_parser_select="error_resilience golomb h264chroma h264dsp h264pred > >>>>h264qpel mpegvideo" > >>>>+h264_parser_select="golomb h264chroma h264dsp h264pred h264qpel > >>>>mpegvideo" > >>>>mpeg4video_parser_select="error_resilience mpegvideo" > >>>>mpegvideo_parser_select="error_resilience mpegvideo" > >>>>vc1_parser_select="mpegvideo" > >>> > >>>This looks suspicious. You're not changing the H.264 parser itself; > >>>was this a transitive dependency? > >> > >>It depends on how you see it I guess. The dependency was in h264.c, > >>which both the h264 decoder and parser use. > > > >That's what I would call a transitive dependency. > > Actually it is hot.
Oh, hotness :) > Say that the h264 decoder consists of two dozen files. Only a few of > these object files are shared between both the parser and the > decoder. (I'm not sure of the exact number of the shared and not > shared files though.) This does not warrant making the h264 decoder > a dependency of the parser - you could just as well make the parser > a dependency of the decoder, if taking things to the extreme. > > Also, semantically, the parser does not depend on the decoder. They > just happen to share some dependencies. > > If you want to split it up properly, you could make the shared > subset a separate entity of its own, like h264_common. But the > parser does not depend on the decoder. We agree 100%; this is just a misunderstanding, a different usage of terms. Diego _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
