Do not think this is really gets used, and it's just more
unnecessary work to keep it in sync esp. for a m-n package.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
the current upstream has been inactive for a long time.
switch to rkta's fork, specifically the oldstable branch which
"contains a very conservative set of bug fixes [potential CVE
fixes] to the last release of tats/w3m."
alpine, freebsd and openbsd has also switched to rkta's fork.
it's likely that other distros will do the same given that w3m
has a bunch of CVEs accumulated (some of which are potentially
fixed in the fork).
in the future we should consider switching over to master branch
instead of oldstable. oldstable is a more lower hanging fruit so
going with that for now.
Ref: https://github.com/tats/w3m/issues/304
Signed-off-by: NRK <nrk@disroot.org>
Closes: https://github.com/gentoo/gentoo/pull/40968
Signed-off-by: Sam James <sam@gentoo.org>
For what it's worth anyway... more of a proof of concept given
ffmpeg-compat-4.4.x likely won't stay around for that long.
Seems to work fine, was able to play opus files despite
USE=-opus through ffmpeg.
This package will still need to be properly handled to avoid
last-rites (does have a few users). I know there are
ffmpeg-6+ patches going around, or could also remove/mask
USE=ffmpeg, but not planning to handle this package myself.
Bug: https://bugs.gentoo.org/834393
Closes: https://github.com/gentoo/gentoo/pull/40942
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Also drop the fdk alternative, it serves no purpose if can
pick the former where it's not set without anything different.
Either way, the || ( lame mp3 ) will be replaced later by just
[lame(-)] which can be done without another revbump.
(not testing this, so not going to add the missing python3.12)
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
The || ( drm libdrm ) will be replaced later by a single
[drm(-)] which can be done without another revbump.
Albeit I wonder if this really needs it set? Seems unusual
to depend on this. Albeit coming from moonlight it wouldn't
be that surprising.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
The || ( lame mp3 ) will be replaced later by a single [lame(-)]
which can be done without another revbump.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Also add :=, not tried but looking at the code this looks
like to be linking with ffmpeg and not just using the tools.
The || ( xml libxml2 ) will be replaced later by just [xml(-)]
which can be done without another revbump.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
...for what it's worth anyway, this depended on an unexisting
>=ffmpeg-9 (USE is masked).
Rather than do the || ( lame mp3 ) dance just depend on lame
directly and drop the bogus >=9. Odds are that when this is
unmasked (if ever), ffmpeg[lame] will have taken over.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Don't really like having to revbump blender over this, but
it's better than users running into conflicts.
The || ( lame mp3 ) will be replaced later by just [lame(-)]
which can be done without another revbump.
... that aside, can some of these versions be dropped?
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
USE=mp3 was renamed to USE=lame but, seeing how this does not have
[encode] and being a game engine, I assume these USE were specified
for mp3 and vorbis playback (not encoding), and then ffmpeg can play
both of these with its builtin decoders regardless of USE.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Do not "think" will need 7.0.x (hard to tell what version is truly LTS
or needed for package support with ffmpeg), can always re-introduce it
as ffmpeg-compat:7 if we ever need it after ffmpeg-8.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Meant to be used with ffmpeg-compat.eclass, see its description.
4.4.5 likely won't see much use at this point if the ffmpeg-5
tracker is not missing anything, but adding anyway for now.
Note that this is not intended to provide libraries for old prebuilt
binaries, it is for building against using the eclass. If prebuilts
want to use this then they will likely need patchelf.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Semi-blind additions, decent deal of these are likely
obsolete, but will be review after old (non-compat)
versions been cleaned.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Both the slotting method and eclass are meant to be as simple
as possible, and isolated so that it does not really need to
work with everything given non-slotted ffmpeg stays.
Did not want turn ffmpeg into a permanent slotting model with
a FFMPEG_SLOT use_expand, eselect, or such potentially turning
it into a special Gentoo-only thing that often need hacks.
Essentially just a way for broken packages to gain time without
blocking everyone's ffmpeg updates.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Similar bare losing a few options, albeit given this version is for
compat and normal users do not need the full set of features (they
should use ffmpeg-6+), skip a few inconvenient IUSE/options:
- chromium: needs a different patch/logic it seems
- glslang: ffmpeg-4 did not have shaderc support, meaning that this
is the only alternative, but this breaks too easily
- mmal: was formerly kept due to lack of IUSE=soc, but would rather
not keep the raspberrypi-userland dependency forever
- npp: not tested, and doubt anyone would want this on anything but
latest ffmpeg (needs nvidia-cuda-toolkit)
- qsv: intelmedia-sdk is dead, and this version did not support libvpl
- amf: likely broken in this old version wrt bug #910271, assuming
newer 4.4.5 did not fix it, but even if it did still little reason
to support things like qsv/amf/nvenc as they are being left behind
in old versions
- soc: plus all its logic, never had a patch for this version anyway
- vmaf: seems broken with current libvmaf version wrt bug #931137,
and old libvmaf has some simd-related issues building (we *could*
patch ffmpeg but at this point I feel little reason to for ffmpeg-4).
- --enable-avresample (bug #722274): off by default, and gone
in later versions, no need to keep it enabled still
- --enable-lto: old version did not take =arg and, either way, keeping
it on this older codebase with a lot of asm just seem like trouble
More would likely fit that list, but not trying to overdo until
something actually causes a problem.
Would've wanted to drop USE=nvenc due to its dependency on old
nv-codec-headers but rather mask it not to have to update revdeps
that depend on nvenc over this.
Hopefully not overlooked anything relevant, spent less time on
this than ffmpeg-7 and "downgrading" an ebuild is awkward.
Ultimately this version is going to be short lived, there is almost
nothing left (that we package) that requires it.
Closes: https://bugs.gentoo.org/722274
Closes: https://bugs.gentoo.org/910271
Closes: https://bugs.gentoo.org/931137
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Almost identical, it only loses a few options compared to live
and need to bring back the __musl__ workaround.
Also switch to a patchset tarball, filesdir was large and would
not want to copy these to media-video/ffmpeg-compat as well.
Additionally include the texinfo-7.2 patch in it wrt bug #947707,
and the missing commit wrt bug #947605 (untested).
Hopefully not overlooked anything relevant, spent less time on
this than ffmpeg-7 and "downgrading" an ebuild is awkward.
Closes: https://bugs.gentoo.org/947605
Closes: https://bugs.gentoo.org/947707
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Using same as the new live ebuild bare for still having yasm
support, and not havng the vulkan-static option yet.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Adds IUSE="npp openmpt rabbitmq", mask where not keyworded (npp
can't gain more keywords, and the others kind of niche and not
really worth it).
soc mask is unneeded given 9999 no longer has the IUSE.
svg (rust) is due to no longer building with old librsvg, it *could*
be patched but this USE is not essential for the average user and
isn't worth maintaining.
There are a lot of pending cleanups due to removed IUSE, but that
has to wait until old versions removal.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>