no comment

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

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

Technically, this is not the "display resolution", since we don't know
what the display will be (zoom and unzoom)

Filip Roséen authored 7 years ago and Marvin Scholz committed 7 years ago

Signed-off-by: Marvin Scholz <epirat07@gmail.com>

Filip Roséen authored 7 years ago and Marvin Scholz committed 7 years ago

It is only being used at this specific location.

Signed-off-by: Marvin Scholz <epirat07@gmail.com>

Filip Roséen authored 7 years ago and Marvin Scholz committed 7 years ago

As the macro is used once, there is no need for it to exist at all.

Signed-off-by: Marvin Scholz <epirat07@gmail.com>

Filip Roséen authored 7 years ago and Marvin Scholz committed 7 years ago

There is no need for FREENULL of the data-members when the whole
object is being deallocated directly after.

Signed-off-by: Marvin Scholz <epirat07@gmail.com>

Filip Roséen authored 7 years ago and Marvin Scholz committed 7 years ago

This change makes it so that a track without <key>Location</key> is
simply ignored, instead of it preventing further tracks within the
playlist to be parsed (as we should of course import as many tracks as
we possibly can, even if there are some we simply cannot handle).

Signed-off-by: Marvin Scholz <epirat07@gmail.com>

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

If the decoder allocates data within video_format_t.p_palette, the
previous implementation would result in a use-after-free (among other
issues) due to the ownership of said p_palette not being well-defined.

This fixes the issue by using video_format_Copy, so that the two fmts
do not refer to the same palette.

fixes: #18334

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

In the affected path we would like to toggle so that the addon is in a
state where it is not enabled by removing that flag (if present), we
would however remove all flags of the addon as the right-hand side
expression will always evaluate to 0.

This fixes the issue by proper use of bitwise negate.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

This will silence a diagnostic related to duplicate const qualifier,
as well as removing the obvious redundance of such qualification.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

The usage of "next" in the second comment is plain wrong, as we are
dealing with a currently playing item, not an upcoming one. The first
comment is redundant.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

If a parent is writable, but has read-only children, the previous
implementation would not delete the children, while deletion of the
parent would go through.

Which would, of course, result in a leak of the children in question.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

This fixes a regression introduced by 1ab80a39, where playback would
not stop if a node was deleted through user-interaction.

As a solution, a new flag to playlist_NodeDeleteExplicit has been
introduced so that we can conditionally stop playback of a deleted
item if desired (there are cases where we do not want such mechanisms,
such as when dealing with deletions of empty nodes).

fixes: #18350

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

Only the core should have permission to do forceful deletion of
entities.

This patch removes a mostly static, from outside of the core, argument
to playlist_NodeDelete, while also making it safer to use as it was
previously possible for anything with access to the function to delete
read-only entities.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

Besides the shortening of the implementation, the refactor also make
sure that we return an error-code if var_SetFloat fails - instead of
silently pretending everything was ok.

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

Filip Roséen authored 7 years ago and Rémi Denis-Courmont committed 7 years ago

The affected translation-unit makes use things from vlc_arrays.h (such
as ARRAY_INIT), but did not directly include it.

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

This saves about 200 bytes per instance.

not uncommon for 1080p h264