GitLab

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

This should handle better ordered chapters and multi-file mkvs
(with --mkv-preload-local-dir)

Should close #3942 #4074 #3810 #4031 #4336 #4501.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

Replace argument = realloc( argument, size ); with realloc_or_free() in modules/*, and while at it add assert( argument ) to mark unhandled ENOMEM conditions, also for malloc().

I tried to split most classes in relevant files.
I regrouped classes when they were very tightened to each other
(in chapters and chapter_command for example )

Note that git doesn't detect correctly the files on this.
git blame -C -C is your friend.

I've had it with this module. There is no support, let alone development, from the matroska folks anymore. I prefer to use ffmpeg which seems to be at least less crash prone and having more of a future. I'm not sure which features we are loosing due to this atm. Likely some meta info, probably ordered chapters and definetly the DVD menu capabilities.

This reverts commit 9b868a2f.

workarounds a bug in KaxBlockVirtual implementation (close #1079).

The check was stupid (always was). This seems way more logical.

- This is REALLY slow.
- It is slow each time you Seek to somewhere that you have not been/played before.
- We ought to create a "Repair index" function instead, like for avi. (no time before release)

This close ticket #1687

This fixes issues with files which had their Cues referenced from this 2nd Seekhead at the end of the file. These files were previously treated as "broken".
A maximum of 10 seek heads is allowed in our module, to prevent recursion. The module is much more flexible now, but not that some Master elements that are allowed to be present multiple times are still handled only once (KaxInfo, KaxTracks, KaxTags).

With these change, most mkv files produced with mkvmerge will produce a significant amount of debug. This is because these files have a load of Junk tags in their eof Seekhead that our parser is warning about. After consulting with robux, it was determined that this was a bug in mkvmerge and a ticket has been filed ( https://www.bunkus.org/bugzilla/show_bug.cgi?id=290 ).

This fixes #1236

mkv: retrieve 2nd seekhead pos, when parsing first seekhead, and add a debug if an Unknown element was seen there.

is too far away (> 500ms) from a key frame.
It will reduce seek precision but the implementation of
ES_OUT_SET_NEXT_DISPLAY_TIME is not good enough yet.

Signed-off-by: Rémi Denis-Courmont <rdenis@simphalempin.com>

Signed-off-by: Rémi Denis-Courmont <rdenis@simphalempin.com>