GitLab

Use OutputStream.last_mux_dts as an approximation of 'current' output
time instead.

This makes it more clear what is this variable for exactly.

Doing anything with that field when we are doing streamcopy (so the
encoder is not opened) should have absolutely no effect.

This more clearly describes what is the variable used for.

Also, move its declaration into the block where it is actually used.

This more clearly describes what is the variable used for.

This makes it more clear what is this variable for exactly.

Currently, if a decoder sets AVFrame.pts, we'd send the same timestamp
to it twice, which is wrong.

Store a variable per OutputStream instead of globals for
audio/video/extradata. This makes the code simpler and cleaner and fixes
2pass with multiple output streams.

Frame duplication now happens in vf_fps.

Strictly speaking it is not correct to call it on refcounted frames, in
any case it is unnecessary, since filtered_frame is always unreferenced
after poll_filter() returns.

It has not been actually used since
37045e42, when the broken vismv code was
removed.

It is private and must not be touched from outside of lavf.

It limits the duration of the data read from a given input.

Using -ss 0 to drop frames with negative timestamps is a perfectly valid
use case.

Some C runtime implementations deadlock when calling threading functions
on the atexit() handler.

Use a simpler wrapper similar to av_log to call the cleanup function
before exit.

Bug-Id: 523

The header is private, and avconv.c doesn't use it
for anything.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

External codec may have corner case reason to fail at init, better
report them instead having the user wonder.

strerror_r is called by av_strerror already.

Fixes a segfault in exit_program() if opening an output file fails.

It is incompatible with refcounted frames and since it's been deprecated
for a long time now, fixing it is not worth the effort.

Hack partially based on a commit by Michael Niedermayer <michaelni@gmx.at>
Should fix (or work around) bug 458.

Integrate the code in the packet reading function, instead of inserting
sleeps in many places.
This is simpler to follow and should work better.