print raw lifetime idents with r# #144897
Conversation
This comment was marked as outdated.
This comment was marked as outdated.
rustbot
added
S-waiting-on-review
|
This PR modifies |
compiler/rustc_span/src/symbol.rs
Outdated
| if let Some(ident) = self.as_raw_lifetime_name_without_apostrophe() { | ||
| write!(f, "'{}", IdentPrinter::new(ident.name, true, None)) | ||
| } else { | ||
| fmt::Display::fmt(&IdentPrinter::new(self.name, self.is_raw_guess(), None), f) | ||
| } |
There was a problem hiding this comment.
Why is this not a change in IdentPrinter instead?
There was a problem hiding this comment.
How do you propose us to make the check in IdentPrinter? It doesn't know about editions because it only accepts the symbol. is_raw_guess and as_raw_lifetime_name_without_apostrophe both need the span edition to know whether some identifier is treated as a keyword or not.
There was a problem hiding this comment.
I suppose it could accept edition as a parameter? otherwise I'm not sure what's even the point of it
There was a problem hiding this comment.
otherwise I'm not sure what's even the point of it
Just looked at the code again. IdentPrinter is used from two general places. One where one has the token data (sorta at the parsing/ast/proc macro stage) in which rawness of idents are known and do not need to be guessed, and one at a later stage where they do. that's why rawness was originally passed as a bool. To improve the design as you highlighted I chose to change the original bool into an enum, but we still have to guess whether a raw lifetime should be printed here and not inside IdentPrinter.
This comment has been minimized.
This comment has been minimized.
fee1-dead
force-pushed
the
raw_lifetimes_printing
branch
2 times, most recently
from
5805f0a to
9705706
Compare
fee1-dead
force-pushed
the
raw_lifetimes_printing
branch
from
9705706 to
b74a97e
Compare
This comment was marked as resolved.
This comment was marked as resolved.
fee1-dead
force-pushed
the
raw_lifetimes_printing
branch
from
b74a97e to
4d45896
Compare
This comment was marked as resolved.
This comment was marked as resolved.
| @@ -3090,7 +3090,13 @@ impl<'a> Parser<'a> { | |||
| if let Some((ident, is_raw)) = self.token.lifetime() { | |||
| // Disallow `'fn`, but with a better error message than `expect_lifetime`. | |||
| if matches!(is_raw, IdentIsRaw::No) && ident.without_first_quote().is_reserved() { | |||
| self.dcx().emit_err(errors::InvalidLabel { span: ident.span, name: ident.name }); | |||
| self.dcx().emit_err(errors::InvalidLabel { | |||
There was a problem hiding this comment.
(alternatively, we could just mimic the diagnostic emitted by expect_lifetime ("lifetimes cannot use keyword names") and just say "labels cannot use keyword names"
| .as_str() | ||
| .strip_prefix("'") | ||
| .expect("only lifetime idents should be passed with RawLifetime mode"); | ||
| Symbol::intern(s) |
There was a problem hiding this comment.
Edit: Disregard, without_first_quote is already interning (in is_raw_lifetime_guess), so it's already interned by this point (I'm not happy about without_first_quote re-interning, seems wasteful in the happy path).
I would just return s.fmt(f); instead of (needlessly) interning the "stripped" lifetime. I don't know if we use IdentPrinter in the happy path at all. If we did this would be a minor perf/mem concern.
compiler/rustc_span/src/symbol.rs
Outdated
| // this should be kept consistent with `Parser::expect_lifetime` found under | ||
| // compiler/rustc_parse/src/parser/ty.rs | ||
| let name_without_apostrophe = self.without_first_quote(); | ||
| name_without_apostrophe.name != self.name | ||
| && ![kw::UnderscoreLifetime, kw::StaticLifetime].contains(&self.name) | ||
| && name_without_apostrophe.is_raw_guess() |
There was a problem hiding this comment.
Since rustc_span actually defines Ident and is_reserved, you could reasonably also define a canonical "is_reserved_lifetime" which you can then reuse.
fmease
added
S-waiting-on-author
fee1-dead
force-pushed
the
raw_lifetimes_printing
branch
from
4d45896 to
f3ff3d3
Compare
|
This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed. Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers. |
This comment has been minimized.
This comment has been minimized.
fee1-dead
force-pushed
the
raw_lifetimes_printing
branch
from
f3ff3d3 to
4970127
Compare
|
@bors r=fmease |
bors
added
S-waiting-on-bors
This replaces #143185 and fixes #143150
cc @fmease