Build failure: folly 2024.11.18 (and therefore watchman) stopped compiling on darwin

[ilyagr]

Steps To Reproduce

Steps to reproduce the behavior:

1. build folly 2024.11.18 (in unstable since a few days ago)

Build log

See https://gist.github.com/ilyagr/6563926a723ed3f6e8dd121e758cd9bd

Additional context

The build chugged along for a couple of minutes, then 2 instances of clang++ took 2 hours at 100% CPU utilization (single core) before failing.

The problem is with some dependency, a few days ago the same version of folly compiled fine. (github:NixOS/nixpkgs/nixpkgs-unstable as of 2024-12-23 20:55 was the last successful time it compiled successfully for me)

I suspect #361878 maybe.

Metadata

 - system: `"aarch64-darwin"`
 - host os: `Darwin 23.6.0, macOS 14.7.2`
 - multi-user?: `yes`
 - sandbox: `no`
 - version: `nix-env (Nix) 2.24.11`
 - nixpkgs: `/nix/store/1jphxmkkqvz69yg0fq6z69dcpfkyy0l9-source`

Notify maintainers

@emilazy
@techknowlogick
@K900

---

Note for maintainers: Please tag this issue in your PR.

---

Add a 👍 reaction to issues you find important.

[paparodeo]

hmm, for some reason I thought that 2024.11.8 didn't have the char_types issue, bummer. a fix + update is already in staging: #362234 but probably will take 2-4 weeks to make it into master.

[paparodeo]

to get master to build can disable tests as it is just the test not compiling which is causing the build to fail.

folly.overrideAttrs { doCheck = false; }

[emilazy]

I think let’s disable the tests on LLVM for now and revert it in staging; will try to put up a PR for that soon if nobody beats me to it. Sorry for being absent when this broke during staging-next or I’d have made sure it got fixed.

[emilazy]

Fixed by #368576.

[ilyagr]

Wonderful, thank you both!