non-compatible racer binary with RUST_SRC_PATH

[ernestrc]

I've started using multirust and suddently racer stopped working for non std completions or go to definitions. I think we should add a warning in the installation instructions about racer-binary/rust-src-path and potentially get racer to throw a warning as well. When invoked from the CLI, it would exit with status 0 in any case.

Any ideas how can we make this incompatibility issue more transparent to consumers of racer?

[phildawes]

racer should work with multirust (that's how I use it). Which versions of everything are you using?
(thanks!)

[ernestrc]

multirust 0.7.0 (b222fcd27 2015-09-24 10:53:38 -0700)
rustc 1.5.0-nightly (130851e03 2015-10-03)
cargo 0.6.0-nightly (7f21e73 2015-10-01)
racer compiled with cargo in las commit (b9750c3)
vim-racer in last commit (dde0d9b557a75858099fd4b873d4a1ed2dfc3111)
arch-linux (also tried on my OSX, same results)

So what I do is use this gist and symlink the nightly toolchain src folder to /usr/local/src/rust/src.

I just realized that I can use go-to-definition only on use std::<mod> or on extern crate <lib1>, on top of the <lib> word. If I try to use it inside one of the nested modules, say for example in use <lib1>, it doesn't work. Is this expected?

To be more specific, I can point you to the repo I'm working on: https://github.com/ernestrc/lbd . If you check it out and open the file src/model/job.rs, using vim-racer, If you try to go-to-definition on top of "fmt" in line 15, it works fine. On the other hand, when I try with "toml" in line 17, it doesn't find anything. If you go to the file src/lbdd.rs and try to do the same on top of "toml" in line 9, it will work. Is there a logical explanation for this?

Thanks mate!

[phildawes]

I think it's because racer doesn't support the cargo '[[bin]]' field yet, so it is not finding src/lbdd.rs as the crate root. I'll add an enhancement issue for this

[phildawes]

created #409

[ernestrc]

Brilliant! Thanks for all the work you've done with Racer, it's going to be an important piece in the future of Rust. Do you think #409 it's doable for a newcomer to the code base?

[phildawes]

Yes, I think it might be. Is quite an isolated piece of work

• add function to cargo.rs to locate the root modules from the cargo.toml
• hook it up to find_possible_crate_root_modules() in nameres.rs

Unfortunately that second function assumes there's only one root in a crate. It might work returning both roots though.