harfbuzz: fix cross compilation

[kira-bruneau]

Motivation for this change

This change indirectly fixes cross compilation for Emacs

Fixes build errors:

• src/meson.build:639:4: ERROR: Problem encountered: Introspection support is requested but it isn't available in cross builds
• Program glib-mkenums mkenums found: NO

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS linux)
• Built on platform(s)
  • NixOS (for armv7l-linux)
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Ensured that relevant documentation is up to date
• Fits CONTRIBUTING.md.

@edolstra

[jtojnar]

Looks like it actually fails on CI due to missing gobject-introspection.

Also I merged #111259, which fixes the introspection error. Though it probably still will not work without #88222.

[kira-bruneau]

Oh oops, I only tried a cross build but didn't test the native build. gobject-introspection is also needed in buildInputs now that strictDeps is enabled.

Tomorrow I'll try to see if it can build with gobject-introspection using the latest version. Thanks! 😄

[kira-bruneau]

Yeah, it won't work without #88222. I'll just close this for now then.

[jtojnar]

Would building without gi, on cross and adding it to buildInputs work? Or not using strictDeps for now?

[kira-bruneau]

Would building without gi, on cross and adding it to buildInputs work?

Yep, it works. I closed this because I was kind of reluctant to keep inputs conditional on cross compilation after you mentioned that they removed the error for cross builds, but I can re-open this and submit the fix.

Then hopefully after #88222 is complete we can remove the isCross conditional.

[Mindavi]

On staging bed8668

Your branch is up to date with 'upstream/staging' (bed86681367c1f57324249db181d142e2efc1f0b).
rick@nixos-asus:~/nixpkgs$ nix build .#pkgsCross.aarch64-multiplatform.harfbuzz
error: builder for '/nix/store/9svk1903cj7llwhzm0jwlxcfq7rkiirk-gtk-doc-1.33.2-aarch64-unknown-linux-gnu.drv' failed with exit code 2;
       last 10 log lines:
       >     p, stdout, stderr = mesonlib.Popen_safe(cmd)
       >   File "/nix/store/pb8yg45560clhm1xqgvrsivfax1p70mq-meson-0.57.1/lib/python3.8/site-packages/mesonbuild/mesonlib/universal.py", line 1320, in Popen_safe
       >     p = subprocess.Popen(args, universal_newlines=True, close_fds=False,
       >   File "/nix/store/sjifaa6ykigz6jxdmgvdnyp36yq7zkbr-python3-3.8.8/lib/python3.8/subprocess.py", line 858, in __init__
       >     self._execute_child(args, executable, preexec_fn, close_fds,
       >   File "/nix/store/sjifaa6ykigz6jxdmgvdnyp36yq7zkbr-python3-3.8.8/lib/python3.8/subprocess.py", line 1596, in _execute_child
       >     self._posix_spawn(args, executable, env, restore_signals,
       >   File "/nix/store/sjifaa6ykigz6jxdmgvdnyp36yq7zkbr-python3-3.8.8/lib/python3.8/subprocess.py", line 1547, in _posix_spawn
       >     self.pid = os.posix_spawn(executable, args, env, **kwargs)
       > OSError: [Errno 8] Exec format error: '/nix/store/h0ic6bhzhnd0ngcskz7739fkh3p392a5-python3-3.8.8-aarch64-unknown-linux-gnu/bin/python3'
       For full logs, run 'nix log /nix/store/9svk1903cj7llwhzm0jwlxcfq7rkiirk-gtk-doc-1.33.2-aarch64-unknown-linux-gnu.drv'.
error: 1 dependencies of derivation '/nix/store/hh2pfkdvfcy7sjj45kxx24j0r46jmmry-glib-2.66.4-aarch64-unknown-linux-gnu.drv' failed to build
error: 1 dependencies of derivation '/nix/store/pyf3dlwz7lv57af158aw5dgbxzya0czb-harfbuzz-2.7.4-aarch64-unknown-linux-gnu.drv' failed to build

Something broken in gtk-doc on staging, I think @SuperSandro2000 saw the same on a PR I made. Or maybe meson itself broken?

I pulled this in for my own tree, and there it builds (possibly something newer on staging broke it). Or my own patches to asciidoc make the difference, that's also possible. So this patch probably works as intended.

This is the patch for asciidoc, maybe it helps? Probably unrelated though.

diff --git a/pkgs/tools/typesetting/asciidoc/default.nix b/pkgs/tools/typesetting/asciidoc/default.nix
index 849da7e606c..95e9f0c7280 100644
--- a/pkgs/tools/typesetting/asciidoc/default.nix
+++ b/pkgs/tools/typesetting/asciidoc/default.nix
@@ -21,6 +21,7 @@
 #, epubcheck ? null
 , gnused ? null
 , coreutils ? null
+, buildPackages
 
 # if true, enable all the below filters and backends
 , enableExtraPlugins ? false
@@ -257,8 +258,8 @@ stdenv.mkDerivation rec {
         -i a2x.py
   '' else ''
     sed -e "s|^ENV =.*|ENV = dict(XML_CATALOG_FILES='${docbook_xml_dtd_45}/xml/dtd/docbook/catalog.xml ${docbook_xsl_ns}/xml/xsl/docbook/catalog.xml ${docbook_xsl}/xml/xsl/docbook/catalog.xml')|" \
-        -e "s|^XSLTPROC =.*|XSLTPROC = '${libxslt.bin}/bin/xsltproc'|" \
-        -e "s|^XMLLINT =.*|XMLLINT = '${libxml2.bin}/bin/xmllint'|" \
+        -e "s|^XSLTPROC =.*|XSLTPROC = '${buildPackages.libxslt.bin}/bin/xsltproc'|" \
+        -e "s|^XMLLINT =.*|XMLLINT = '${buildPackages.libxml2.bin}/bin/xmllint'|" \
         -i a2x.py
   '') + ''
     patchShebangs .

[SuperSandro2000]

Something broken in gtk-doc on staging

It executes the python for the target platform on the build platform.

Or my own patches to asciidoc make the difference, that's also possible.

That shouldn't make any difference because the error is in gtk-doc.

I think we need to tell gtk-doc to use the python of the build platform with some flag but I am unsure how and which. @jtojnar can you help?

[sternenseemann]

Done in #123712.