lp:~mpv-team/mpv/master

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

40535. By Ricardo Constantino on 2018-03-15

DOCS/options: clarify that --end also supports relative time

40534. By Philip Sequeira <email address hidden> on 2018-03-15

stream_file: enable cache for FUSE filesystems on OS X

Requested in #634. Better late than never?

40533. By Philip Sequeira <email address hidden> on 2018-03-15

stream_file: add more network file systems (Linux)

Fixes #5643.

40532. By wm4 <wm4@nowhere> on 2018-03-15

client API: send MPV_EVENT_SHUTDOWN only once

Before this change, mpv_wait_event() could inconsistently return
multiple MPV_EVENT_SHUTDOWN events to a single mpv_handle, up to the
point of spamming the event queue under certain circumstances. Change
this and just send it exactly once to each mpv_handle.

Some client API users might have weird requirements about destroying
their state asynchronously (and not reacting immediately to the SHUTDOWN
event). This change will help a bit to make this less weird and
surprising.

40531. By wm4 <wm4@nowhere> on 2018-03-15

all: replace mpv_detach_destroy() with mpv_destroy()

40530. By wm4 <wm4@nowhere> on 2018-03-15

client API: rename mpv_detach_destroy() to mpv_destroy()

Since this has clearer semantics now, the old name is just clunky and
confusing.

40529. By wm4 <wm4@nowhere> on 2018-03-15

client API: add mpv_create_weak_client()

40528. By wm4 <wm4@nowhere> on 2018-03-15

client API: cleanup mpv_handle termination

This changes how mpv_terminate_destroy() and mpv_detach_destroy()
behave. The doxygen in client.h tries to point out the differences. The
goal is to make this more useful to the API user (making it behave like
refcounting).

This will be refined in follow up commits.

Initialization is unfortunately closely tied to termination, so that
changes as well. This also removes earlier hacks that make sure that
some parts of FFmpeg initialization are run in the playback thread
(instead of the user's thread). This does not matter with standard
FFmpeg, and I have no reason to care about this anymore.

40527. By wm4 <wm4@nowhere> on 2018-03-15

player: move locale check to mp_create()

The intention is to reduce annoying differences between mpv CLI and
libmpv, and there's no reason to have the locale check only in libmpv
(although it doesn't help with any real issues either).

40526. By wm4 <wm4@nowhere> on 2018-03-15

player: shuffle around CLI exit code handling

Remove the weird prepare_exit_cplayer() function, and fold the contents
into mpv_main() and mp_initialize().