Perl uses GNU extensions if they are available. Its build system
tries to detect whether the libc is one that recognizes _GNU_SOURCE
with bad heuristics, that don't work for e.g. riscv64 musl. Upstream
is not willing to unconditionally set _GNU_SOURCE like programs
usually do, because of an IMO misguided fear of somebody one day
writing a libc that interprets _GNU_SOURCE to mean something
different, so the only thing we can really do is set it ourselves.
On libcs that don't gate GNU extensions behind _GNU_SOURCE, like BSDs,
this should be a no-op. This should also improve things for cross
where musl builds are currently broken due to perl-cross setting
_GNU_SOURCE for the configure step but not the build step, although I
haven't tested that.
Link: https://github.com/Perl/perl5/issues/16807#issuecomment-544099623
Link: https://github.com/arsv/perl-cross/issues/158
The perl package started depending on `apple-sdk` via references in
`lib/perl5/5.40.0/darwin-thread-multi-2level/Config_heavy.pl`:
```
fd --type=file . /nix/store/cwlch1mxd40ayx8qy7hl8q2scrwgqhg6-perl-5.40.0/lib/ -X rg "/nix/store/lsjl29pwp5if71jfgxlv8fifsrpax805-apple-sdk-11.3" {}
/nix/store/cwlch1mxd40ayx8qy7hl8q2scrwgqhg6-perl-5.40.0/lib/perl5/5.40.0/darwin-thread-multi-2level/Config_heavy.pl
1039:incpth='/nix/store/j3dhkx4m14m38drh3f40bccrwrgzlsg5-libcxx-19.1.7-dev/include /nix/store/hklang3njvw0f5mgra9vsvn3c2zi2v45-compiler-rt-libc-19.1.7-dev/include /nix/store/g4qcalcnvm5n393xq8p9swxd8rc0xxzj-libxcrypt-4.4.38/include /nix/store/f4z2y0zn2bmrwywp67y3h21cmb6ky547-libiconv-109-dev/include /nix/store/17a9rq4bw8drbhlr3dijrmxzfqlz8z85-libresolv-83-dev/include /nix/store/34bb5prqghfl0hpindhk8xlc5s2q2xjg-libsbuf-14.1.0-dev/include /nix/store/vzi6s9yh1i12bkj9abyfp43y6x1fdzpl-cups-headers-2.4.11/include /no-such-path/resource-root/include /nix/store/lsjl29pwp5if71jfgxlv8fifsrpax805-apple-sdk-11.3/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include /no-such-path/include'
1087:libpth='/nix/store/g4qcalcnvm5n393xq8p9swxd8rc0xxzj-libxcrypt-4.4.38/lib /no-such-path/resource-root/lib /nix/store/lsjl29pwp5if71jfgxlv8fifsrpax805-apple-sdk-11.3/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/lib /no-such-path/lib'
1092:libspath=' /nix/store/g4qcalcnvm5n393xq8p9swxd8rc0xxzj-libxcrypt-4.4.38/lib /no-such-path/resource-root/lib /nix/store/lsjl29pwp5if71jfgxlv8fifsrpax805-apple-sdk-11.3/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/lib /no-such-path/lib'
1331:timeincl='/nix/store/lsjl29pwp5if71jfgxlv8fifsrpax805-apple-sdk-11.3/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include/sys/time.h '
```
This change increases the closure size of `perl` from ~115MiB to ~1.11GiB
on darwin systems.
`apple-sdk` seems to be referenced via `stdenv.cc.fallback_sdk`.
Following the exisitng pattern in
`pkgs/development/interpreters/perl/interpreter.nix`,
that removes removes references to `stdenv.cc` and `stdenv.cc.cc`,
this commit also removes the reference to
`apple-sdk` / `stdenv.cc.fallback_sdk`.
After final improvements to the official formatter implementation,
this commit now performs the first treewide reformat of Nix files using it.
This is part of the implementation of RFC 166.
Only "inactive" files are reformatted, meaning only files that
aren't being touched by any PR with activity in the past 2 months.
This is to avoid conflicts for PRs that might soon be merged.
Later we can do a full treewide reformat to get the rest,
which should not cause as many conflicts.
A CI check has already been running for some time to ensure that new and
already-formatted files are formatted, so the files being reformatted here
should also stay formatted.
This commit was automatically created and can be verified using
nix-build https://github.com/infinisil/treewide-nixpkgs-reformat-script/archive/a08b3a4d199c6124ac5b36a889d9099b4383463f.tar.gz \
--argstr baseRev b32a094368
result/bin/apply-formatting $NIXPKGS_PATH
In preparation for the deprecation of `stdenv.isX`.
These shorthands are not conducive to cross-compilation because they
hide the platforms.
Darwin might get cross-compilation for which the continued usage of `stdenv.isDarwin` will get in the way
One example of why this is bad and especially affects compiler packages
https://www.github.com/NixOS/nixpkgs/pull/343059
There are too many files to go through manually but a treewide should
get users thinking when they see a `hostPlatform.isX` in a place where it
doesn't make sense.
```
fd --type f "\.nix" | xargs sd --fixed-strings "stdenv.is" "stdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "stdenv'.is" "stdenv'.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "clangStdenv.is" "clangStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "gccStdenv.is" "gccStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "stdenvNoCC.is" "stdenvNoCC.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "inherit (stdenv) is" "inherit (stdenv.hostPlatform) is"
fd --type f "\.nix" | xargs sd --fixed-strings "buildStdenv.is" "buildStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "effectiveStdenv.is" "effectiveStdenv.hostPlatform.is"
fd --type f "\.nix" | xargs sd --fixed-strings "originalStdenv.is" "originalStdenv.hostPlatform.is"
```
Without the change `nixpkgs` exposes unevaluatable attribute as:
$ nix build --no-link -f. perlInterpreters.perl536.perlOnBuild
error: attribute 'perl536' missing
at pkgs/development/interpreters/perl/interpreter.nix:207:37:
206| perlOnBuildForBuild = override pkgsBuildBuild.${perlAttr};
207| perlOnBuildForHost = override pkgsBuildHost.${perlAttr};
| ^
208| perlOnBuildForTarget = override pkgsBuildTarget.${perlAttr};
Did you mean one of perl538 or perl540?
The interpreter depends on presence of `perl536` attribute which wa
removed recently.
While preparing this change, I read the git blame on all of the files I
touched. I saw a working lifetime of building this system which we use
every day and love dearly and keep maintained ourselves. I saw commits
from a 14 year range between 2003 to 2017!! I could not be more thankful
for Eelco's work on building large parts of the foundation of nixpkgs
that all of us rely on now.
However, the end date of that range of the files I looked at the blame
on was 2017. I did not see surviving code from any newer date than that.
Looking at the Git logs, Eelco has been working on other things, and
that's totally fine.
However, it means that our maintenance metadata is out of date on a lot
of packages, and *that*'s the reason I am submitting this change. There
are a lot of packages that don't have anyone with their name on them to
be pinged if they need attention, even if they have had recent activity
(although it is never clear if recent activity was just someone fixing
it because ZHF or because the package actually matters to them).
There are a lot of packages with storied history that maybe don't need
to be in the set anymore at all since they have not been touched in
years; or maybe they are simply finished.
Empty maintainer lists should be a sign that we need to figure out who
maintains it or potentially remove it if it has rotted, and allowing the
maintainer list to be empty if it is already not maintained is part of a
healthy repository ecology.
Either way, I would like to have the maintenance metadata not mislead
anyone into sending Eelco emails about packages he doesn't, in practice,
work on anymore. I have not removed his name from everything; there are
some things that he is the upstream for or has worked on more recently,
for instance, like Nix, which I have left alone.
The "correct" solution to this problem (makeBinaryWrapper fails to be
able to compile the binary wrapper when building a freebsd cross env) is
that makeBinaryWrapper should propagate or embed a reference to the
compiler. However, this causes build failures on musl, so this will have
to do for now.
enableParallelBuilding=false was inadvertently set in f4162143, this
commit reverts to the previous behavior of disabling parallel builds
only during during cross compilation.
The change was made due to problems with local builds during the
upgrade to perl 5.38.0, and should not have been included in the commit.
Stig Palmquist