| Age | Commit message (Collapse) | Author |
|---|
|
In October, Google released version 2018.0 of `shaderc` (see https://github.com/google/shaderc/releases), and now AUR has `shaderc` package which builds v2018.0 (https://aur.archlinux.org/packages/shaderc/). `shaderc-git` package has also updated so that it provides `shaderc`.
|
|
This is always expected to be running on Arch or similarly configured
systems so we know exactly where the system's Python is going to be, and
that's the one we want to use. No need to risk problems from e.g. custom
builds ahead of it in the path.
|
|
Accidentally changed in last commit.
|
|
This allows building without network, caching the waf download even when
doing a clean build, and is cleaner semantically in general.
We still run mpv's bootstrap script, but we now tell it not to attempt
the download. It will simply check that the waf version it expects is
present. If that check fails, it will bail and the PKGBUILD will have to
be updated.
|
|
It was previously in the main pacman package.
|
|
|
|
This fixes a build failure that several users have experienced due to a
weird package situation. See the comment+link in PKGBUILD for details.
|
|
Upstream has dropped it and is back to supporting upstream ffmpeg by
default.
|
|
Building against system ffmpeg is still possible by uncommenting a line
in the PKGBUILD. In that case we're now using the new
--enable-ffmpeg-upstream flag, so we can drop the patch.
While we're in the neighborhood, remove that old samba include stuff. I
guess it was needed at some point; pkg-config handles it now.
|
|
Can't just revert the commit anymore, as the surrounding code has
changed since. Add a patch with a simple workaround that will hopefully
not break too often.
PR #5033 has been merged(ish) upstream. Remove the patch.
mpv now requires git master ffmpeg; the latest release is not new
enough. This package probably should start depending on ffmpeg-git, but
I'm not switching it yet because that isn't provided by the current
ffmpeg-mpv-git package.
We'll just live with the _dirty in the version for now. If this goes on,
I'll strip it from the pkgver, but it will stay in the built binary as
long as it's true, to respect those handling bug reports upstream. Some
kind of option to avoid applying the patch when using ffmpeg-mpv should
be doable, though, so those who are following upstream's recommendations
can get a fully supported build without -dirty.
|
|
Also apply a patch for the issue that caused this mess. See the comments
on the AUR page for some links to what's going on.
This is a temporary situation until things settle down. I'm not sure
what I'll do if mpv continues to reject upstream ffmpeg long-term, but I
probably won't be continuing to carry these patches.
|
|
|
|
I don't really think this is necessary; mpv's releases are pretty
arbitrary and users of this package should just update whenever. But
bumping isn't any harder than complaining about semantics in the
comments when it's flagged so I'll just do it. Not sure if I'll continue
to do it in the future.
|
|
|
|
|
|
So we don't need the option.
|
|
|
|
With some minor changes.
I still don't really like doing it this way... Maybe we'll switch to a
date-based version or something in the future.
|
|
The functionality it provides is already possible with user shaders:
https://github.com/bjin/mpv-prescalers
(I'm doing this now because I'm going out of town soon and don't want the
package to break if upstream removes the flag while I'm gone.)
|
|
|
|
|
|
|
|
Some of them are already installed by mpv's own installer, ending up with
duplicates. The official package only removed the dups, but I'm just removing
all of them and leaving it up to mpv what it wants to install.
|
|
|
|
Releases aren't relevant anymore (at least for now). The new pkgver
format matches the official package in community. Hopefully this will
reduce confusion.
|
|
Also, use a better (?) way to avoid the false fakeroot dependency, and
print detected dependencies when packaging.
|
|
|
|
|