Hang in wasmer during item collection

[matklad]

λ git clone https://github.com/matklad/wasmer-repro && cd wasmer-repro
Cloning into 'wasmer-repro'...
remote: Enumerating objects: 12, done.
remote: Counting objects: 100% (12/12), done.
remote: Compressing objects: 100% (7/7), done.
remote: Total 12 (delta 0), reused 12 (delta 0), pack-reused 0
Receiving objects: 100% (12/12), 7.08 KiB | 7.08 MiB/s, done.

13:21:38|~/tmp/wasmer-repro|master✓
λ rust-analyzer analysis-stats .
Database loaded:     966.26ms, 155minstr

For a change, this is a hang not in chalk. Perhaps related to nameres changes? Didn't try to minimize yet

[matklad]

Good commit: 1f7d2a6
Bad commit: eb08a27

Seems like this might be the attributes name res chage? @jonas-schievink

[matklad]

Fun fact: this was to be a repro for a different bug in a different project: wasmerio/wasmer#2328

[jonas-schievink]

764241e is still good, probably caused by #8898

[jonas-schievink]

Repro:

#[unresolved]
extern "C" {
    #[unresolved]
    fn f();
}

[jonas-schievink]

This is from js-sys, actually, which was reported as having the same bug in #8414 (comment)

[jonas-schievink]

I have a suspicion that the ItemTree represents this as

#[unresolved]
#[unresolved]
fn f(); // IS_IN_EXTERN_BLOCK

...and this results in problems further down the line. We need to represent extern blocks as a first-class item in the ItemTree in order to allow attribute macros on them anyways, so I'll do that and see what happens.

Also, having a way to pretty-print ItemTrees in a Rust-like syntax would be handy to debug this.

[matklad]

Also, having a way to pretty-print ItemTrees in a Rust-like syntax would be handy to debug this.

Two easy steps for this:

• step 1: implement automated code formatter
• step 2: use ast::make to conveniently convert ItemTree back to (pretty) ast

[jonas-schievink]

Hmm, it still hangs while indexing wasmer_compiler