[MPlayer-dev-eng] [PATCH] fix length in asf files

[Reimar Döffinger]

On Tue, Nov 10, 2009 at 10:07:48AM -0500, compn wrote:
> asfheader-force_best_stream.diff

Will probably break mms in many cases, since the "best" stream
this selects might not actually be available.

> avi_check_idxflags.diff

A sample would be nice. Particularly since id is not set at that point.
Looks to me like this patch just ends up disabling that check
completely.

> codecs_conf-workarounds.diff
> configure-static.diff

Some custom hacks I'd say.

> demux_avi-forceidx_fallback.diff

Seems pointless to me. Always using -forceidx is not that bad for local
files, and for remote files you don't want it automatically.
Switching default to lavf for AVI probably is a better solution in
the long term.

> demux_lavf-add_dv_mts_preferred.diff

DV is probably a good idea, auto-detection is fairly reliable for it now
I think.
MPEG-TS has a few issues with feature-differences, I think FFmpeg one
still doesn't handle e.g. EVOBs well, native one does not (yet?) support
DVB-teletext (or any other kind of subtitle).

> demux_lavf-ID_VIDEO_ROTATION.diff

Interesting, but too specific I think.

> demux_lavf-probe_small_files.diff

Applied something slightly less ugly - though further improvements
are still possible here, in particular increasing the probe buffer
if the detection was not possible or reliable.

> demux_mpg-short_video_fix.diff

Probably to "hackish"/too much code for a corner case for us to really
care about.

> libavformat_mov-avoid_reporting_empty_av_streams.diff

Not sure about that. Might be better to leave it to the using
application to determine that kind of thing.
Normally duration should allow detecting empty streams and filtering
them out.

> libavformat_mov-track_rotation_metadata.diff

Better solutions might be in progress already (not sure I admit).

> mencoder-ext_file_conf.diff

without the code duplication and with load_per_protocol_config it would
be nice I think.

> mencoder_01_identify.diff

I don't think this is really a must, most users would have mplayer
installed, too, and can use that.

> mencoder_02_audio_only_hack.diff

Well... Not that bad, but would be better combined with some
refactoring.

> mencoder_03_joint_av_eof.diff

Ugh, I don't think we should risk calling muxer_write_chunk with len <=
0.
Apart from that I am not sure which behaviour is more desirable: stop
encoding when both streams end or as soon as one ends...

> mencoder_04_av_sync_hack.diff

Likely this just covers up a rather big bug somewhere else.

> mencoder_05_mov_lavf_pts_var_fps.diff
> mencoder_06_asf_pts_passing.diff
> mencoder_07_demux_update_pts.diff
> mencoder_09_rebase_pts_when_seeking.diff
> mencoder_10_correct_pts.diff

Looks like a partial correct-pts implementation for mencoder.
Not so great for us I think, even without the needless code duplication.

> mencoder_08_decoded_before_dups.diff

That looks interesting. Not sure if mux_v->size>0 is the best condition
though, that would depend on when exactly there is an issue without it.

> mov-ignore_reference_trak.diff
> mov_01_edl_kf_search.diff
> mov_02_mean_framerate.diff
> mov_03_moof_fragments.diff
> mov_04_mvhd_length.diff
> mov_05_ffmpeg_bitrate.diff
> mov_06_tkhd_matrix_scale.diff
> mov_07_increase_sample_pts_size.diff

No idea, but mov is preferredly handled by lavf anyway.

> mp3lib-increase_maxframesize.diff
> mp3lib-validate_winarray_index.diff

Should go upstream.

> mpegvideo-revert_r18381.diff

That definitely reintroduces a bug, if it causes issues somewhere it
would be a very good idea to discuss with FFmpeg...

> reduce_spurious_logging.diff

Might be worth looking through.

> stream_file-clear_eof_on_seek.diff

I guess that applies to all protocols and probably should be handled in
stream_seek_long.

> vd_ffmpeg-restrict_dimension_correction.diff

No idea at all. Part of the patch is for lavf though...

> vd_qtvideo-validate_ImageDesc_size.diff

yes, unfortunately lavf does not give us everything necessary to use QT
binary decoders :-(

> ve_raw-stride_support.diff

Call me stupid, but I fail to see the point?!

> vf_vo-stride_support.diff

draw_frame is deprecated, so the right solution is update any vo still
using it :-)