[stable] Backport fix for #60989 "ICE with incorrect turbofish"

[oli-obk]

stable fix backport for #60989

do not merge yet, untested

[rust-highfive]

r? @petrochenkov

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

[rust-highfive]

⚠️ Warning ⚠️

• Pull requests are usually filed against the master branch for this repo, but this one is against stable. Please double check that you specified the right target!

[petrochenkov]

r? @pietroalbini

[eddyb]

Started stage1 test build to help out, apparently LLVM builds make it hard to test stable backports.

EDIT: most of it took like 7 seconds, with all the TableGen stuff maybe about a minute.
I guess I had an old enough LLVM build in ccache, this machine is not that fast.

[eddyb]

error: internal compiler error: src/librustc/hir/def.rs:271: attempted .def_id() on invalid def: Local(NodeId(13246))

I talked to @oli-obk, and he doesn't mind me taking over (since I already have a build dir for this).

[emilyalbini]

This is ok from a release point of view, but I'm not knowledgeable enough to be the assignee for this.

[eddyb]

./x.py test --stage 1 src/test/ui succeeds now for me, what's left?
Presumably someone other than me has to review it (@varkor?).

[emilyalbini]

Anyone from @rust-lang/compiler up for a review of this?

[varkor]

@bors r+ p=1000

[bors]

📌 Commit 6e81f82 has been approved by varkor

[bors]

⌛ Testing commit 6e81f82 with merge e6bb10d673b9ef4119e5f9de26bd4dd2a138b8c3...

[cuviper]

There's no such label, but let's not forget this for master/nightly too...

[varkor]

@cuviper: #60989 isn't being marked as fix, so we have the issue to remind us 👍

[bors]

💔 Test failed - status-appveyor

[cuviper]

I don't know if this is spurious, but the rls test failure is here:
https://ci.appveyor.com/project/rust-lang/rust/builds/24765767/job/c72eopevht8wxajk#L10885

thread 'client_changing_workspace_lib_retains_diagnostics' panicked at 'assertion failed: bin.diagnostics[0].message.contains("mismatched types\n\nexpected u32, found u64")', src\tools\rls\tests\client.rs:275:5

[varkor]

@bors retry

[varkor]

(I'm also not sure, but this is blocking the stable release, so I'm going to hope it was.)

[bors]

⌛ Testing commit 6e81f82 with merge 6c8dd30950da3cc2a89b3e211e1ac1743c67421c...

[eddyb]

cc @Xanewok for #61085 (comment)

[emilyalbini]

Yeah that should be spurious.

[emilyalbini]

Unfortunately we can't include this into 1.35.0: the spurious failure moved the tentative release time to around 2AM in my timezone, and y'all don't want me doing a release that late. @eddyb or @oli-obk, can any of you rebase this PR on master and beta-nominated+stable-nominated it?

Closing this PR.

@bors r- retry

[eddyb]

Note for whoever wants to rebase this on master (before I have a chance to look at it): Def was renamed to Res, so at the very least there's that, but also rewrite_self_ctor might have changed its signature and may be more aggressive now. (if so, it should be changed to return just Res)