Move binder and polarity parsing into parse_generic_ty_bound
#127103
Conversation
| #![feature(precise_capturing)] | ||
|
|
||
| fn polarity() -> impl Sized + ?use<> {} | ||
| //~^ ERROR expected identifier, found keyword `use` |
There was a problem hiding this comment.
I chose against generalizing error_lt_bound_with_modifiers for use<> precise capturing syntax. I guess I could? Do you think anyone would really ever write ?use<> or for<'a> use<'a>??
| let modifiers = self.parse_trait_bound_modifiers()?; | ||
| let (mut lifetime_defs, binder_span) = self.parse_late_bound_lifetime_defs()?; |
There was a problem hiding this comment.
This should make the diff of #127054 really simple -- just invert these two lines, fixup some comments, and bless tests.
| = note: for more information on higher-ranked polymorphism, visit https://doc.rust-lang.org/nomicon/hrtb.html | ||
| help: consider making the bound lifetime-generic with a new `'a` lifetime | ||
| | | ||
| LL | fn foo<T>() where for<'a> T: for<'a> 'a {} |
There was a problem hiding this comment.
lol this kinda sucks tho -- maybe i should bail the parser or something so we don't get to resolution here.
There was a problem hiding this comment.
Yeah, I guess we could do that. Don't think it's super important. Can be addressed at some other point(tm)
| #![feature(precise_capturing)] | ||
|
|
||
| fn polarity() -> impl Sized + ?use<> {} | ||
| //~^ ERROR expected identifier, found keyword `use` |
| if self.token.is_lifetime() { | ||
| let _: ErrorGuaranteed = self.error_lt_bound_with_modifiers(modifiers, binder_span); | ||
| return self.parse_generic_lt_bound(lo, has_parens); | ||
| } |
There was a problem hiding this comment.
Does this regress the following valid code (semantically valid 2015–2018, syntactically valid >2018):
fn f() where for<'a> 'a + std::fmt::Debug: {}If so, I remember that there are some helper methods with which you can "look ahead for trailing a +"
There was a problem hiding this comment.
Ah, no. The for<> belongs to the predicate, not to the type. So it should be fine and likely doesn't regress it.
There was a problem hiding this comment.
Yeah, and fn f() where (for<'a> 'a + std::fmt::Debug): {} is not valid
|
I checked and that code you shared does not regress @bors r=fmease |
…iaskrgr Rollup of 9 pull requests Successful merges: - rust-lang#126822 (Bootstrap command refactoring: port more `Command` usages to `BootstrapCmd` (step 2)) - rust-lang#126835 (Simplifications in match lowering) - rust-lang#126953 (std: separate TLS key creation from TLS access) - rust-lang#127045 (Rename `super_predicates_of` and similar queries to `explicit_*` to note that they're not elaborated) - rust-lang#127075 (rustc_data_structures: Explicitly check for 64-bit atomics support) - rust-lang#127101 (remove redundant match statement from dataflow const prop) - rust-lang#127102 (Rename fuchsia builder and bump Fuchsia) - rust-lang#127103 (Move binder and polarity parsing into `parse_generic_ty_bound`) - rust-lang#127108 (unify `dylib` and `bin_helpers` and create `shared_helpers::parse_value_from_args`) r? `@ghost` `@rustbot` modify labels: rollup
Rollup merge of rust-lang#127103 - compiler-errors:tighten-trait-bound-parsing, r=fmease Move binder and polarity parsing into `parse_generic_ty_bound` Let's pull out the parts of rust-lang#127054 which just: 1. Make the parsing code less confusing 2. Fix `?use<>` (to correctly be denied) 3. Improve `T: for<'a> 'a` diagnostics This should have no user-facing effects on stable parsing. r? fmease
Let's pull out the parts of #127054 which just:
?use<>(to correctly be denied)T: for<'a> 'adiagnosticsThis should have no user-facing effects on stable parsing.
r? fmease