* Removed dependencies that are only used
for configuration of Calamares in case
anyone wants to use extras.
Another package can be created and used
in order to install Gentoo via GUI.
* Removed tests that are changing
host variables, and testing directly
on host, unsandbox'ed.
Signed-off-by: Stefan Cristian B. <stefan.cristian+git@rogentos.ro>
Closes: https://github.com/gentoo/gentoo/pull/41021
Signed-off-by: Viorel Munteanu <ceamac@gentoo.org>
As the ebuild's ewarn notes, --gpu-api=opengl still works with USE=egl
through egl-x11/wayland, this only drops the deprecated/legacy gl-x11.
Debated renaming USE=egl to USE=opengl instead given it's more commonly
used, but given we've been calling it deprecated in metadata.xml and
that mpv options are all egl-* it may come as confusing.
The ewarn is not overly essential given we enable USE=egl by default,
but mpv tend to have a few minimalist users that disable everything
they can. I may add that if do this, may want to consider USE=vulkan
instead if it works well given mpv upstream favours it by default now.
Untested but tentatively add aqua for -Dgl (formerly controlled
through USE=opengl), believe gl-cocoa is always available? Please
report if this causes issues.
Closes: https://bugs.gentoo.org/953107
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
All the configuration is jammed into Makefile fragments with the
expectation that you edit it by hand. Prime 1989 technology. Capitulate
to this demand, but use a slightly better editor (echo).
Closes: https://bugs.gentoo.org/725428
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
The software is extremely crusty. As mentioned in the previous commit,
it hasn't been maintained since 1993 and there are derivative projects
which likely solve these issues. The code has "worked" since 1993 and I
suspect unlikely to ever be of much interest to new platforms, given
that it's only even useful as a point of reference compared to its
modern derivatives. Keep in mind the code is so old that it's from back
when the most recent brand new C revision supported implicit function
declarations.
Trying to solve its C issues seems unworthwhile. Just -std=c89 it.
Actually, gnu89 since it depends on `#define linux` to detect the
platform. :P
Closes: https://bugs.gentoo.org/919301
Closes: https://bugs.gentoo.org/944870
Closes: https://bugs.gentoo.org/945068
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
The makefiles for this project are from 1989, back when software
developers still knew how to write real build systems. For example, real
build systems are intelligent enough to:
- mark loads of rules as "-command", which tells make to not fail if the
command fails
- run make by concatenating makefile fragments and piping it to `make -f -`
- recursively run more makes via concatenation
- pass -k to all make invocations, so that we try to build as much as
possible even if we can't succeed in the end
- mandatorily require running make from a shell script that sets up
everything, somehow
There is only one slight problem with this otherwise brilliant design.
Namely, none of it is a good idea.
ANYWAY, this software does not build due to Modern C violations. It is
also entirely uninteresting to invest serious time in the build system
for this thing, or even attempt to report it, since the software exists
as a wikipedia "historically significant events" page plus random
third-party download sites and web.archive.org and a progression of
modernized and extended forks. The actual project was declared finished
and done with in 1993.
I'm told keeping the package around is important because, as the father
of an entire industry, it's important to have as a point of reference
whenever doing various kinds of work. Still, the darned thing could do a
bit better about actually compiling. Step 1 is to error out on compile
failures, rather than succeeding and then failing to run the built
executables / have the input files for "dobin" and erroring out with
inscrutable messages.
Signed-off-by: Eli Schwartz <eschwartz@gentoo.org>
6.9-3 -> 6.9-4 changes:
Updated:
- musl-no-execinfo.patch
As usual only tend to test musl (+clang/libcxx) with .0 release
and there was one minor build failure due to a new execinfo-related
function not being guarded by #if/#endif.
Runtime also seems fine on llvm-musl, qutebrowser-9999 test pass
and can browse normally as well.
Still not including the other patches given they (hopefully)
shouldn't be that long lived.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Almost identical to 4.0.0 but this release clears a up a lot of
packaging headaches thus the ebuild simplifications.
Signed-off-by: Ionen Wolkens <ionen@gentoo.org>
Stefan Cristian B.
Viorel Munteanu
Yongxiang Liang
Jim Tupper
Andreas Sturmlechner
Thomas Beierlein
Ionen Wolkens
Arthur Zamarin
Hans de Graaff
Eli Schwartz
Michael Orlitzky
Sam James
Alexey Shvetsov