Wesley Moore

rustc tests appear to assume unix-style paths and line endings: https://github.com/rust-lang/rust/blob/master/src/test/ui/module-macro_use-arguments.stderr and are normalised as such in the test runner: https://github.com/rust-lang/rust/blob/44628f7273052d0bb8e8218518dacab210e1fe0d/src/tools/compiletest/src/runtest.rs#L3595-L3620

In the latter commits I reworked how `BufferError` is handled to allow errors like this to be generated: ``` error: unexpected end of buffer ┌─ formats/opentype.fathom:34:18 │ 34 │ magic

Noticed a couple of ways to handle this in a URLO forum thread: 1. [Spawn a thread to do the work with a custom stack size](https://users.rust-lang.org/t/what-can-i-do-to-avoid-thread-main-has-overflowed-its-stack-when-working-with-large-arrays/77091/3?u=wezm) 1. [Increase the stack...

Yep 👍

I'm basing the makedepends on what's done for the [ripgrep](https://git.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/ripgrep), which is a rust tool in the community repos. It depends on cargo, which is specified as a package provided...

It sounds like you have rustup installed without a default toolchain set. I had wondered about this. At the moment it think the PKGBUILD is using whatever `cargo` it finds...

I pushed a revised `PKGBUILD` that uses the absolute path to `cargo`, give that a try and see if that helps.

Interesting. I suppose the NeovimGtk one should have a unique name. I wonder what the consequence of having two versions of the same shim in the runtime path. I suppose...

Ok I'll have a look into changing the package to use or mirror the Makefile behviour.

I updated the `PKGBUILD` to better mirror the Makefile.