`everybody_loops` is no more

[netzdoktor]

I just build cargo-breaking from the source, and it failed to run on a very simple crate. The error is:

Caused by:
       Compiling XXXXX v1.3.1 (/XXXX)
    error: argument to `unpretty` must be one of `normal`, `identified`, `expanded`, `expanded,identified`, `expanded,hygiene`, `ast-tree`, `ast-tree,expanded`, `hir`, `hir,identified`, `hir,typed`, `hir-tree`, `thir-tree`, `mir` or `mir-cfg`; got everybody_loops
    
    error: could not compile `XXXXX` due to previous error

The root cause is here (in a call to cargo):

cargo-breaking/src/glue.rs

Line 20 in 56dafaf

.args(&["-Z", "unpretty=everybody_loops"])

In rust-lang/rust#93913, this option was removed. Pull request rust-lang/rust#94669 suggests that thir-tree is a replacement. When I changed that locally in my cargo-breaking build, it started working again.

[o0Ignition0o]

Hey! Great catch, it is indeed not up to date!

Would you mind opening a pull request? I'd be happy to accept it!

[o0Ignition0o]

Edit:

It looks like #42 would work out of the box, since it doesn't use the same approach.

We're going to land this instead imo.

It's worth noting we're considering a switch to cargo-semver-checks which builds on top of a very powerful and interesting query engine.

[netzdoktor]

Well, I saw that there was a talk on cargo-breaking at EuroRust (mentioning rustdoc in the abstract). I wanted to try it out but then realized that you had this issue and the #42-related branch.

However, on that branch, I am also getting strange errors:

Error: Failed run rustdoc

Caused by:
    0: Failed to deserialize rustdoc output
    1: missing field `param_names` at line 1 column 1304

Shall I open a separate issue for that or is #42 anyhow still "in the making"?

I am anyhow not in a hurry. I am teaching students about APIs and SemVer and thought about adding this handy tool to my presentation.