Rustc shim is now a wrapper

[albertlarsan68]

This is preparatory work to make clippy work in the repo (#107628).
This changes the rustc shim in bootstrap to be RUSTC_WRAPPER, instead of RUSTC.

[rustbot]

r? @ozkanonur

(rustbot has picked a reviewer for you, use r? to override)

[jyn514]

@albertlarsan68 locally you have rustc in PATH. In CI, you must always set RUSTC or use the absolute path to the rustc binary, or it will fail.

[the8472]

I think this would also break setups using sccache via RUSTC_WRAPPER.

[rust-log-analyzer]

The job x86_64-gnu-tools failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

Caused by:
  process didn't exit successfully: `/checkout/obj/build/x86_64-unknown-linux-gnu/stage0-std/release/build/libc-d2d75a169efdb097/build-script-build` (exit status: 101)
  --- stdout
  cargo:rerun-if-changed=build.rs
  --- stderr
  --- stderr
  thread 'main' panicked at 'Failed to get rustc version', /cargo/registry/src/index.crates.io-6f17d22bba15001f/libc-0.2.140/build.rs:188:9
warning: build failed, waiting for other jobs to finish...
Build completed unsuccessfully in 0:00:20
cat: /tmp/toolstate/toolstates.json: No such file or directory

[jyn514]

I think this would also break setups using sccache via RUSTC_WRAPPER.

"break" is a little strong, they just won't be cached, right? How often is it possible to cache these in practice anyway, won't you have to rebuild every time you rebase (or even more frequently for stage 1 builds)?

[jyn514]

That said, we could probably do this for check/clippy only if it causes issues.

[the8472]

sccache sometimes saves me quite a bit of compile time when when switching back and forth between commits or blasting away parts of the build dir.

[albertlarsan68]

I could "cache" the RUSTC_WRAPPER, to call him with the right command and flags.

[bors]

☔ The latest upstream changes (presumably #111231) made this pull request unmergeable. Please resolve the merge conflicts.