cmake: add check-pc-files hook to check broken pc files

[Artturin]

see #172150

test by building seexpr

Description of changes

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 22.05 Release Notes (or backporting 21.11 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
  • (Release notes changes) Ran nixos/doc/manual/md-to-db.sh to update generated release notes
• Fits CONTRIBUTING.md.

[Artturin]

@alexshpilkin please create a issue here with instructions on how to fix and then i'll link to it instead of the pr

[SuperSandro2000]

@ofborg eval

[veprbl]

Can we put everything in the main setup hook? It's much easier to browse all the stdenv magic in one predictable place.

[Artturin]

we can but i want to keep them separate

I've renamed the hook to check-pc-files-hook.sh

[veprbl]

Why?

[SuperSandro2000]

because it is a hook and hooks mostly end in -hook.sh

[veprbl]

That's not the question. Why @Artturin wants a separate file?

[Artturin]

Its a separate hook and there's no reason to keep them in the same file

all hooks can be found with fd --type f "hook.sh"

[trofi]

Bisect claims b1149dc cmake: add check-pc-files hook to check broken pc files broke glslang build on staging:

$ nix build -f. -L glslang
...
glslang> checking for references to /build/ in /nix/store/a7g64d1ghjr5qxvrvmp11dna9cf29nvs-glslang-1.3.224.1...
glslang> Broken paths found in a .pc file! /nix/store/a7g64d1ghjr5qxvrvmp11dna9cf29nvs-glslang-1.3.224.1/lib/pkgconfig/SPIRV-Tools.pc
glslang> The following lines have issues (specifically '//' in paths).
glslang> 3:libdir=${prefix}//nix/store/a7g64d1ghjr5qxvrvmp11dna9cf29nvs-glslang-1.3.224.1/lib
glslang> 4:includedir=${prefix}//nix/store/a7g64d1ghjr5qxvrvmp11dna9cf29nvs-glslang-1.3.224.1/include
glslang> It is very likely that paths are being joined improperly.
glslang> ex: "${prefix}/@CMAKE_INSTALL_LIBDIR@" should be "@CMAKE_INSTALL_FULL_LIBDIR@"
glslang> Please see https://github.com/NixOS/nixpkgs/issues/144170 for more details.

[alexshpilkin]

@Artturin @trofi

Right, I’m sorry for falling of the face of the Earth and thank you for picking up the work, but that is why the .pc file of a vendored library is relevant: glslang bundles (an older version of) spirv-headers and installs its .pc file, which consequently trips this check. Per my notes, the same applies to shaderc. Both need the same patch as in spirv-headers applied somewhere inside their source tree.

My notes also show gromacs and valhalla as unfixed, but I honestly don’t remember what was difficult to fix about them. That should be it (as of May, don’t know about now).

[jtojnar]

Then that is a bug in the build script – vendored dependencies should not install .pc files.

[vcunat]

So I at least unblocked glslang by fixing those files in a1bad74. The libraries are installed, so at a quick glance I can't see if they're used by another package. Certainly feel free to improve this; overall the contents of output didn't seem very... "clean".

[vcunat]

This was quite impactful. I fixed this in more than a dozen packages myself and others fixed some, too.

[Artturin]

Also check .cmake files #247474