The syntax is ${parameter:-word} (i.e. previously this used
"latestTag" instead of the actual value).
(Fixes a regression from #85278.)
Also: Even though getting the latest tag isn't really security critical
(as long as Git itself is secure against untrusted input), I'd prefer to
switch from the Git to the HTTPS protocol (for authentication of the
server and encryption + uses a standard port).
This caused none of these flags to have any effect. That's because
the configure command looked like this:
./configure --prefix=/nix/store/svhl0fjdj1jl2sqcppy5vnzpfi4gj3d3-gbsplay-2016-12-17 \
--without-test\ --without-contrib\ --disable-devdsp\ --enable-pulse\ --disable-alsa\ --disable-midi\ --disable-nas\ --disable-dsound\ --disable-i18n
with one giant flag '--without-test --without-contrib...', containing
internal spaces.
This can be seen in `nix log nixpkgs.gbsplay`, in this line:
configure flags: --prefix=/nix/store/svhl0fjdj1jl2sqcppy5vnzpfi4gj3d3-gbsplay-2016-12-17 --without-test\ --without-contrib\ --disable-devdsp\ --enable-pulse\ --disable-alsa\ --disable-midi\ --disable-nas\ --disable-dsound\ --disable-i18n
and then in the fact that features like "devdsp" and "midi" are listed
as enabled in later output, and source files like plugout_midi.c are
included in the build.
I don't have a real opinion on whether it's better to have these flags
or not, but it's clear the author's intention was to pass them. So,
fix the attr name so they get passed.
patchelf seems to generate an ELF that "strip" cannot process:
strip: /tmp/stNGAZvd: not enough room for program headers, try linking with -N
strip:/tmp/stNGAZvd[.interp]: bad value
(This unfortunately does not cause any errors or even error output in
the build log, since the default strip hook redirects them to /dev/null
and ignores failures.)
The lack of stripping ends up leaking paths to the compiler in the debug
symbols. This more than doubles the transitive closure size of
alacritty.
Impact on closure size: 314.5M -> 131.8M.
Context: discussion in https://github.com/NixOS/nixpkgs/pull/82630
Mesa has been supporting S3TC natively without requiring these libraries
since the S3TC patent expired in December 2017.
Following nh2's advice, I want to be more involved in nixpkgs,
and I'll start to review changes to programs that are important to me,
and that didn't have a long list of maintainers yet.
