halide: 18.0.0 -> 19.0.0

[twesterhout]

Updates halide and python3Packages.halide to the newest version. See upstream release notes for more info.

On the Nix side, I added two patches that fix CMake packaging issues (without them installPhase fails).

Including ${llvmPackages.clang}/resource-root/include fixes compilation issues that @FliegendeWurst was having in #371394.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• 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/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 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
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[twesterhout]

Added wasmSupport flag that conditionally adds wabt. Also fixed some issues with exploding runtime closure size---now it's back to ~200MB, yay.

[twesterhout]

If darwin runners don't get to building the derivation until tomorrow, I can probably get access to an aarch64-darwin system locally...

[FliegendeWurst]

Both regular and wasmSupport = true build for me, x64 Linux.

If you checked that the package works I can merge.

[FliegendeWurst]

My understanding is that ofborg currently doesn't do Darwin. I always use the nix-community builder for testing.

[twesterhout]

Thanks @FliegendeWurst! It seems to build on aarch64-darwin as well, so I think it's ready to be merged.

[twesterhout]

Yay, built on x86_64-darwin too 🥳

[GaetanLepage]

I'm seeing the following failure on darwin:

The following tests FAILED:
         30 - mullapudi2016_reorder (Failed)                    autoschedulers_cpu mullapudi2016 multithreaded
Errors while running CTest
FAILED: CMakeFiles/test.util

Are you building with the sandbox enabled ?

[twesterhout]

Here's the build log for darwin: https://github.com/twesterhout/quick-ci-experiments/actions/runs/12888827983/job/35934639711

Are you building with the sandbox enabled ?

I think so. It's the default isn't it?

Do you happen to have any more info about the failure? Just to see whether it's a legitimate failure or something weird in the test (in which case we should just disable this test).

[twesterhout]

Thanks for the suggestions @natsukium! All fixed.

[GaetanLepage]

Do you happen to have any more info about the failure? Just to see whether it's a legitimate failure or something weird in the test (in which case we should just disable this test).

Unfortunately, there doesn't seem to be any more context...

[GaetanLepage]

@ofborg build halide, halide.tests

[GaetanLepage]

Actually, I missed the error:

 42/724 Test  #31: mullapudi2016_reorder ......................................***Failed  Required regular expression not found. Regex=[Success!
]  3.68 sec
Autoscheduler time (2) is slower than expected:
======================
Manual time: 7.72782ms
Auto time: 17.3505ms
======================

So it looks like a flaky test because it can fail if the builder is too slow.
Maybe we should simply skip it...