The current postInstall removes all of $out/share, which also includes pulse's
vapi/deps files. Certain projects (budgie-desktop, for example) require these
definitions, but do not require the full pulseaudio server. Preserving these
files allows these projects to only depend on libpulseaudio.
Fixes#73463.
The x86_64 and aarch64 SDK attributes are named differently and don't
have the same contents. This was the motivation to try dropping the SDK
dependency altogether, relying only on the Libc and some frameworks.
continuation of #109595
pkgconfig was aliased in 2018, however, it remained in
all-packages.nix due to its wide usage. This cleans
up the remaining references to pkgs.pkgsconfig and
moves the entry to aliases.nix.
python3Packages.pkgconfig remained unchanged because
it's the canonical name of the upstream package
on pypi.
When cross-compiling, pulseaudio seems to not find some m4 macro
providing GSETTINGS_RULES.
However, apart from the obviously missing gsettings support, this works
just fine.
It only increases the closure size by 0.5M and users who do not set
the NixOS option `hardware.pulseaudio.package = pkgs.pulseaudioFull;`
will be stumped by their bluetooth audio not working.
This one was a bit tricky to find because it only causes a runtime
error, and pulseaudio has shims for most ALSA UCM methods except one.
(I guess nobody ever really tests pulseaudio in combination with
UCM-less ALSA?)
ALSA 1.1.8 had ${alsa-dev}/include/alsa/ in the pkg-config file, which
was considered wrong and fixed in 1.1.9.
However, pulseaudio was relying on being able to include ALSA headers
like <asoundlib.h> and <use-case.h> rather than <alsa/asoundlib.h> and
<alsa/use-case.h>. (For asoundlib.h it only causes a warning, because
the ALSA guys created a shim for that header.)
These two patches change pulseaudio to use the correct include
directives.
He prefers to contribute to his own nixpkgs fork triton.
Since he is still marked as maintainer in many packages
this leaves the wrong impression he still maintains those.
The most complex problems were from dealing with switches reverted in
the meantime (gcc5, gmp6, ncurses6).
It's likely that darwin is (still) broken nontrivially.
