dcpomatic/src/lib/ffmpeg_decoder.cc, branch debunDCP-o-matic DCP toolshttps://git.carlh.net/cgit/dcpomatic/atom?h=debun2026-03-12T00:24:06ZUse dts rather than pts when noting the time of frames we are ignoring.2026-03-12T00:24:06ZCarl Hetheringtoncth@carlh.net2026-03-12T00:02:51Zurn:sha1:4285ff59480d526161905ecdc5fc05b57be7ed58
The docs for av_read_frame() say
"pkt->pts can be AV_NOPTS_VALUE if the video format has B-frames, so it
is better to rely on pkt->dts if you do not decompress the payload."
Before this there could be long hangs while we go through many packets
without pts values.
Add a debug note.2026-03-12T00:24:06ZCarl Hetheringtoncth@carlh.net2026-03-12T00:02:18Zurn:sha1:ef526ea3d6d6196f0e38ac9770504877388fc47aHandle FFmpeg content with only subtitles enabled better.2026-02-21T01:37:42ZCarl Hetheringtoncth@carlh.net2026-02-21T01:37:42Zurn:sha1:c2adc362e4ce0a64d7e7a1c6b634b024d862c8c1
Previously if some FFmpeg content was set to use only its subtitles, and
the first subtitle was 1 minute in to the content, the content would be
selected for pass() until its first minute had been scanned.
This could be slow, and incorrect as on the way to this 1 minute we want
to look at other content.
Here we store the PTS of dropped packets to use as a fallback position
when there is no video or audio enabled.
Fix error when seeking a file with no video and >1 audio stream (#3128).2025-12-14T23:20:53ZCarl Hetheringtoncth@carlh.net2025-12-12T10:43:45Zurn:sha1:52b3eee4e7285e26469b2a5f4d9d9eac20aba387Try AVSEEK_FLAG_ANY if _BACKWARD fails.2025-12-08T19:50:53ZCarl Hetheringtoncth@carlh.net2025-12-07T01:17:34Zurn:sha1:9bd8b3304f35f0b27bf18f2167302f1fd5d3f92b
We usually want to do _BACKWARD as this should seek to the keyframe
before the seek position. On some files this fails, and then it
seems that _ANY is a good next thing to try.
Clarify and improve decoder/player/butler logging a bit.2025-12-06T10:12:31ZCarl Hetheringtoncth@carlh.net2025-12-06T10:11:47Zurn:sha1:5029bf180a8b29c8e9f1aeef1c398b05e01eb663Cleanup: can now remove the _NC log variants.2025-12-05T20:36:53ZCarl Hetheringtoncth@carlh.net2025-12-05T20:36:53Zurn:sha1:59e358ae25f84314afc6d8235aecaec124822f3fDisable use of stream IDs if there are duplicates, rather than rewriting.2025-10-15T20:33:26ZCarl Hetheringtoncth@carlh.net2025-10-14T15:02:15Zurn:sha1:2d8d05c2e7ad67ebac2ff250670a219a891d09ca
It turns out that FFmpeg decoders (e.g. flv, see FFmpeg
25faaa311a74efdfdc4fed56996d7338ed807488) check stream IDs and sometimes
create new streams if they see one that they didn't see before. If we
change stream IDs we break this.
Here we try to use stream indices in cases where the IDs are duplicated.
We also account for the case where a new stream appears during
examination. This wasn't covered by tests until the FFmpeg commit
mentioned above, were the flv decoder creates a new stream during
examination of boon_telly.mkv.
Remove compose.hpp includes.2025-07-10T18:50:32ZCarl Hetheringtoncth@carlh.net2025-07-09T20:58:37Zurn:sha1:bad01065584014aee4024e7cbc4a66acac7617e1
sed -i "/include.*compose.hpp/d;" src/lib/*.cc src/wx/*.cc src/wx/*.h src/tools/*.cc src/lib/*.h test/*.cc
Replace String::compose with fmt.2025-07-10T18:50:32ZCarl Hetheringtoncth@carlh.net2025-07-09T20:53:27Zurn:sha1:62c34b28567a097e8f22576e7d7891bd3dbe0ac0
sed -i "/Plural-Forms/n;/%100/n;/scanf/n;s/%[123456789]/{}/g" src/lib/*.cc src/lib/*.h src/wx/*.cc src/tools/*.cc src/lib/po/*.po src/wx/po/*.po src/tools/po/*.po test/*.cc
sed -i "s/String::compose */fmt::format/g" src/lib/*.cc src/lib/*.h src/wx/*.cc src/tools/*.cc test/*.cc