On 06/12/2016 01:57 PM, Carl Eugen Hoyos wrote: > DeHackEd <git <at> dehacked.net> writes: > >>> So the bug is that the file gets longer when read with FFmpeg? >> >> Heh... that's my fault for copy/pasting from the wrong invocation. >> I have two test cases, one generated from the above testsrc and >> one that actually consists of a 28+ hour recording from a real TV >> source. It's a "real world" scenario but the file is about 90 >> gigabytes. > > I am both able to record arbitrarily long mpegts files and to run > FFmpeg to produce long samples. > > But my question is: What goes wrong with the file? > And how can I reproduce this with a self-produced (or self- > recorded) file?
The effects vary a bit. One guy in IRC basically had ffmpeg crash for him every 26.5 hours when he was saving to HLS - I don't have his actual output on me right now. I have an libav* application fail because av_write_frame() returns an error when the DTS moves backwards. Based on chatting with other people in #ffmpeg-devel it seemed like it was best to implement this correction into the demuxers themselves. Now it seems fixing libavformat/utils.c user functions may be the better way to go based on this thread so far. > > Carl Eugen > > _______________________________________________ > ffmpeg-devel mailing list > ffmpeg-devel@ffmpeg.org > http://ffmpeg.org/mailman/listinfo/ffmpeg-devel > _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel