Give correct suggestion for a typo in raw pointers #145903

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Open

Kivooeo wants to merge 1 commit into rust-lang:master from Kivooeo:c-style-pointer

+556 −0

Member

Kivooeo commented Aug 26, 2025 •

edited

Loading

This adds a check for when user wrote const* T instead of *const T, I just saw how a C-person made this typo and compiler suggests this

 --> src/main.rs:2:17
  |
2 |     let p: const* u8 = 0 as _;
  |                 ^
  |
help: add `mut` or `const` here
  |
2 |     let p: const*mut  u8 = 0 as _;
  |                  +++
2 |     let p: const*const  u8 = 0 as _;
  |                  +++++

which is very incorrect

in my current approach we don't recover from this because otherwise we will get a wall from errors which is very verbose and 0 helpful

also fixes #136602

r? compiler


          add parser check for pointer typo

18a0d8e

rustbot assigned wesleywiser

rustbot added S-waiting-on-review T-compiler labels

compiler-errors requested changes

View reviewed changes

compiler/rustc_parse/src/parser/ty.rs

    
                              return Ok(self.mk_ty(kw_span.to(star_span), TyKind::Err(guar)));

                          }

                      }

Member

compiler-errors Aug 27, 2025

Suggested change

compiler/rustc_parse/src/parser/ty.rs

    
                                  )

                                  .emit();

                              return Ok(self.mk_ty(kw_span.to(star_span), TyKind::Err(guar)));

Member

compiler-errors Aug 27, 2025

Why not recover this like a TyKind::RawPtr? Recovering like TyKind::Err seems to be strictly worse.

compiler/rustc_parse/src/parser/ty.rs

    
                          let kw_span = self.token.span;

                          self.bump();

                          if self.eat(exp!(Star)) {

Member

compiler-errors Aug 27, 2025

This should try to parse a subsequent type via a snapshot. Or at least check that lookahead(2) can begin a type

compiler/rustc_parse/src/parser/ty.rs

    
                          let kw_span = self.token.span;

                          self.bump();

                          if self.eat(exp!(Star)) {

Member

compiler-errors Aug 27, 2025 •

edited

Loading

This eat is always true, since we checked the lookahead above

compiler/rustc_parse/src/parser/ty.rs

    
                                  .dcx()

                                  .struct_span_err(

                                      kw_span,

                                      "raw pointer type must have a star before the mutability keyword",

Member

compiler-errors Aug 27, 2025 •

edited

Loading

Suggested change

      
                                    "raw pointer type must have a star before the mutability keyword",
          
                                    "raw pointer types are written like `*{...} T`",

I find the phrasing kind of awkward -- "must have a star before the mutability keyword" doesn't really point to the fact that there is a star, and it's just misplaced. And also I wouldn't call "const" to be a mutability keyword.

I'm not sure if what I suggested is any better, but I wonder if there's a way to suggest the change here that sounds more descriptive.

compiler/rustc_parse/src/parser/ty.rs

    
            @@ -276,6 +276,34 @@ impl<'a> Parser<'a> {
          
                  ) -> PResult<'a, Box<Ty>> {

                      let allow_qpath_recovery = recover_qpath == RecoverQPath::Yes;

                      maybe_recover_from_interpolated_ty_qpath!(self, allow_qpath_recovery);

                      if (self.token.is_keyword(kw::Const) || self.token.is_keyword(kw::Mut))

Member

compiler-errors Aug 27, 2025

Suggested change

      
                    if (self.token.is_keyword(kw::Const) || self.token.is_keyword(kw::Mut))
          
                    // Comment that describes what we're trying to do here.
          
                    if (self.token.is_keyword(kw::Const) || self.token.is_keyword(kw::Mut))

compiler/rustc_parse/src/parser/ty.rs

    
            @@ -276,6 +276,34 @@ impl<'a> Parser<'a> {
          
                  ) -> PResult<'a, Box<Ty>> {

                      let allow_qpath_recovery = recover_qpath == RecoverQPath::Yes;

                      maybe_recover_from_interpolated_ty_qpath!(self, allow_qpath_recovery);

                      if (self.token.is_keyword(kw::Const) || self.token.is_keyword(kw::Mut))

Member

compiler-errors Aug 27, 2025

This probably should also be moved into a separate function and also moved way down into the actually if else chain below rather than being hit first on every call to parse_ty_common

compiler-errors added S-waiting-on-author and removed S-waiting-on-review labels

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

S-waiting-on-author T-compiler