Rustup to rustc 1.36.0-nightly (1764b2972 2019-05-12)

[Manishearth]

No description provided.

[oli-obk]

I have time right now, want me to take over?

[mati865]

Related PR rust-lang/rust#60630

[oli-obk]

It should be mostly paint by numbers by now (if not or unclear otherwise, just leave it for me for tomorrow).

If any symbol is missing, just add it to the list in utils::sym. If any string literal has a type error, change it to sym::string_content. If that's not possible because it's not a valid identifier, find a new name for it and use your_valid_identifier: "random string" in utils::sym, if you get weird type errors when using a symbol, you are probably missing a * (due to lazy static), so use *sym::foo everywhere. There's still one weird thing with a method that I didn't look too closely at (one of the hack methods with loads of collect calls in it). Also gazillions of unused imports, just remove them with rustfix in the end.

[Manishearth]

We have a problem, cross referential statics doesn't really work, you get lots of temporary lifetime issues.

Perhaps we can just manually initialize these with unsafe?

Also we should move all our usage of match_def_path over to just having a clippy::diagnostic::foobar tag on items in the stdlib.

[Manishearth]

I think what we'll need to do is to reserve a nice block of a thousand symbol ids in rustc, and then have a proc macro generate them.

[Manishearth]

I ... I just had a super devious idea. DefIds don't change for a given compiler, right?

We could totally add an intrinsic get_def_id::<T>() that gives you a DefId, and use that in match_def_path. Then we can just match_def_path::<Option>(def)

kinda prefer just annotating things, though.

[oli-obk]

@bors r+

[bors]

📌 Commit dfbc74b has been approved by oli-obk

[bors]

⌛ Testing commit dfbc74b with merge ad3269c...

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: oli-obk
Pushing ad3269c to master...