Allow running x test reference locally

[jyn514]

Previously this would break for me with:

Testing stage2 mdbook src/doc/reference (aarch64-apple-darwin)
error: manifest path `mdbook-spec/Cargo.toml` does not exist

but then would say "build completed successfully" ... I guess that's related to toolstate somehow.

I have no idea how this ever passed in CI.

[rustbot]

r? @clubby789

rustbot has assigned @clubby789.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[jyn514]

maybe related: the CI job prints this every time it tests the reference:

##[group]Testing stage2 mdbook src/doc/reference (x86_64-unknown-linux-gnu)
[2025-12-03T00:03:32Z WARN  mdbook::preprocess::cmd] The command wasn't found, is the "spec" preprocessor installed?
[2025-12-03T00:03:32Z WARN  mdbook::preprocess::cmd] 	Command: cargo run --release --manifest-path mdbook-spec/Cargo.toml

[ehuss]

The error message is normal (and can safely be ignored), and will be removed in a future update.

Can you say more about what issues you ran into, or why these changes seem needed? The way rustbook works is that it overrides the preprocessors. We actually don't want the cargo run to succeed, since it would then be using the wrong workspace.

[jyn514]

hm, ok. i ran into this because the reference failed mysteriously on a PR, and then mysteriously fixed itself again. see #149565 (comment)

[ehuss]

There was a problem that the toolstate mechanism broke, which caused some problems. Part of that is a timing issue in that toolstate becomes locked down during "beta week" (the week before branch). #149577 fixed the immediate problem, and you just need to rebase on latest main to resolve that. (#149612 fixes the alerting/toolstate problem that led up to that.)

[jyn514]

gotcha, thank you!