On Thu, Nov 14, 2013 at 08:46:40PM -0500, Ronald S. Bultje wrote: > Hi, > > On Thu, Nov 14, 2013 at 7:58 PM, Luca Barbato <[email protected]> wrote: > > > From: "Ronald S. Bultje" <[email protected]> > > > > Originally written by Ronald S. Bultje <[email protected]> and > > Clément Bœsch <[email protected]> > > > > Further contributions by: > > > Please define the "further contributions". I am the original author, > together with Clement, I'd like to know what you changed. You're throwing a > blob at me without individual author's contribution, I don't think that's > appropriate. >
It seems this message was ignored and the decoder just pushed. Since it
seems the changes were not done openly, I'm going to summarize what I see
at first glance here for the record:
- [C] variable and function renames (ctx → avctx, sbits_inv →
bits_with_sign, res → ret, f → cur_frame, lflvl → lflvl_ptr...)
- [C] spaces, line breaks, ...
- [C] use of typedef VP9Filter instead of struct
- Comment changes:
-// for some reason the sign bit is at the end, not the start, of a bit sequence
+// The sign bit is at the end, not the start, of a bit sequence
- clipping added in update_prob() (can you elaborate? who did that? why?)
- typo fix in av_log message ("intialize")
- checks for refidx[N]->buf[0] instead of refidx[N] only
- moved large chunks of code outside the file (i haven't looked at the
differences in those)
- tile size type from unsigned to int64_t (can you elaborate? who did
that? why?)
- simplify bw adaptivity changed to a single memcpy
- replace av_frame_ref for !invisible with an unref for invisible case
- demacrotization in vp9_decode_packet()
- alloc references instead of frame buffers in vp9_decode_init()
Some of these changes will be backported in the repository from where this
was picked. Unfortunately, due to the lack of authorship attributions,
people who made those changes won't have correct/accurate credits in the
backports. If someone wants to claim authorship on some of these changes
(or some other), please do it.
Also, as one of the author/contributor to VP9, just like what Ronald
stated, I'm interested about the "why" of some of these changes.
Note: this forked decoder is outdated in comparison to the original
decoder, and lacks some recent changes. Of course, given how much you
tried to diverge from the original code, it might be harder to pick them.
--
Clément B.
pgpAAQFa_ShAK.pgp
Description: PGP signature
_______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
