Modernize diagnostic for indeterminate trait object lifetime bounds #152226
Conversation
fmease
commented
Feb 6, 2026
fmease
commented
- remove suggestion from the diagnostic message (bad style, too long) and turn it into a structured suggestion
- replace object type with trait object type since the former is an outdated term
rustbot
added
S-waiting-on-review
|
r? @chenyukang rustbot has assigned @chenyukang. Use Why was this reviewer chosen?The reviewer was selected based on:
|
| // FIXME: Account for trailing plus `dyn Trait+`, the need of parens in | ||
| // `*const dyn Trait` and `Fn() -> *const dyn Trait`. |
There was a problem hiding this comment.
I've decided against fixing these ~minor issues in this PR because it's a rabbit hole I didn't want to pursue at the moment:
Most diagnostics in the compiler that suggest adding + /* bound */ to a type don't consider both last cases mentioned in the FIXME above (and I bet none properly account for trailing pluses1). E.g., suggest_new_region_bound (see FIXME).
We do have some helpers in place for cases like this like hir::Generics::bounds_span_for_suggestions but it's not general enough (e.g., I can't use it here) & it doesn't account for everything (like raw pointers).
I hope this paints a clear enough picture. I'd rather fix a bunch of these diagnostic suggestions at once in a separate future PR by providing better HIR helpers for these cases. I'm already tracking this in my internal issue tracker (and I might upstream this issue to r-l/r if I won't get to write a PR soon-ish). From a technical standpoint, it shouldn't be hard at all, you just need to check the parent & the child of the HIR type. I just really want to avoid duplicating code for this.
Footnotes
-
Since trailing pluses aren't represented in the AST, HIR etc. and you can't just trim &
contains('+')-check the relevant part of the source text as obtained viasource_map()since it wouldn't account for source code comments; you'd actually need to re-lex the snippet (which is overkill for such a niche issue) ↩
|
This PR conflicts with #152076 which I'm assigned to and which I'll hopefully review today. Can't tell yet which one's more likely to get merged first. |
|
@bors r=chenyukang |
rust-bors
bot
added
S-waiting-on-bors
…ect-lifetime-diag, r=chenyukang Modernize diagnostic for indeterminate trait object lifetime bounds * remove suggestion from the diagnostic message (bad style, too long) and turn it into a structured suggestion * replace *object type* with *trait object type* since the former is an outdated term
Rollup of 7 pull requests Successful merges: - #151152 (Add FCW for derive helper attributes that will conflict with built-in attributes) - #151954 (Add help message suggesting explicit reference cast for From/TryFrom) - #152148 (Move `impl Interner for TyCtxt` to its own submodule) - #152226 (Modernize diagnostic for indeterminate trait object lifetime bounds) - #150688 (typeck: Make it clearer that `check_pat_lit` only handles literal patterns) - #152293 (Format heterogeneous try blocks) - #152396 (Uplift `Predicate::allow_normalization` to `rustc_type_ir`)
…ect-lifetime-diag, r=chenyukang Modernize diagnostic for indeterminate trait object lifetime bounds * remove suggestion from the diagnostic message (bad style, too long) and turn it into a structured suggestion * replace *object type* with *trait object type* since the former is an outdated term
…ect-lifetime-diag, r=chenyukang Modernize diagnostic for indeterminate trait object lifetime bounds * remove suggestion from the diagnostic message (bad style, too long) and turn it into a structured suggestion * replace *object type* with *trait object type* since the former is an outdated term
…ect-lifetime-diag, r=chenyukang Modernize diagnostic for indeterminate trait object lifetime bounds * remove suggestion from the diagnostic message (bad style, too long) and turn it into a structured suggestion * replace *object type* with *trait object type* since the former is an outdated term
…uwer Rollup of 11 pull requests Successful merges: - #152364 (Port a lot of attributes to the new parser) - #151954 (Add help message suggesting explicit reference cast for From/TryFrom) - #152148 (Move `impl Interner for TyCtxt` to its own submodule) - #152226 (Modernize diagnostic for indeterminate trait object lifetime bounds) - #152351 (Remove `SubdiagMessage` in favour of the identical `DiagMessage`) - #152417 (Move the needs-drop check for `arena_cache` queries out of macro code) - #150688 (typeck: Make it clearer that `check_pat_lit` only handles literal patterns) - #152293 (Format heterogeneous try blocks) - #152355 (Update documentation of rustc_macros) - #152396 (Uplift `Predicate::allow_normalization` to `rustc_type_ir`) - #152425 (Port #![test_runner] to the attribute parser)
…uwer Rollup of 10 pull requests Successful merges: - #152364 (Port a lot of attributes to the new parser) - #151954 (Add help message suggesting explicit reference cast for From/TryFrom) - #152148 (Move `impl Interner for TyCtxt` to its own submodule) - #152226 (Modernize diagnostic for indeterminate trait object lifetime bounds) - #152351 (Remove `SubdiagMessage` in favour of the identical `DiagMessage`) - #152417 (Move the needs-drop check for `arena_cache` queries out of macro code) - #150688 (typeck: Make it clearer that `check_pat_lit` only handles literal patterns) - #152293 (Format heterogeneous try blocks) - #152355 (Update documentation of rustc_macros) - #152396 (Uplift `Predicate::allow_normalization` to `rustc_type_ir`)
3 participants
