Piotr Osiewicz

Oh, right, it looks like I should've updated the weak/strong match threshold as well. I've actually removed the distinction and the fuzzy score should now be prefered to LSP score....

It looks like we always treat `file://` paths as URLs, which is why we struggle with colons for rows/columns

Oddly enough, this does not happen with Rust. We do not go into this endless update loop with Rust files.

Yeah, tree-sitter has nothing to do with it. All I'm saying is that it might be an issue with our integration with specific language servers, as RA does not have...

Ah, I see; thanks for pointing that issue out. That's definitely another good argument for not squeezing every last bit out of `cargo clean -p` if it's gonna be reworked...

Thanks for a nice repro! While we could try to guess the full path based on it's suffix, I wonder how we could handle cases where there are multiple paths...

Does it happen for all Rust projects for you? Or just some specific one? Also re: > should only re-compile the update files rather than all the code It is...

I'm not sure if that's caused by Zed; could you post your settings.json? Maybe there's something in there that's causing this (I'm mostly thinking of RA initialization_options).

There's an "expand macro recursively" action for Rust files. Does that fit your use case?

That's done as of #9284; I've checked it locally but just in case anybody wanted to double-check, signed artifacts are available in e.g: https://github.com/zed-industries/zed/actions/runs/8297127089