-
Notifications
You must be signed in to change notification settings - Fork 12.9k
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.
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
Move E0101 and E0102 logic into new E0282 mechanism #40013 #41236
Conversation
r? @pnkfelix (rust_highfive has picked a reviewer for you, use r? to override) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let me know what you think. If what I want is unclear, I can also try to do it locally, since I think it ought to be fairly simple and I'm not sure if I explained it well.
@@ -65,24 +65,41 @@ impl<'a, 'gcx, 'tcx> TraitErrorKey<'tcx> { | |||
struct FindLocalByTypeVisitor<'a, 'gcx: 'a + 'tcx, 'tcx: 'a> { | |||
infcx: &'a InferCtxt<'a, 'gcx, 'tcx>, | |||
target_ty: &'a Ty<'tcx>, | |||
found_pattern: Option<&'a Pat>, | |||
hir_map: &'a hir::map::Map<'gcx>, | |||
found_pattern: Option<&'gcx Pat>, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
maybe we should call this found_local_pattern
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
ok
@@ -835,42 +854,51 @@ impl<'a, 'gcx, 'tcx> InferCtxt<'a, 'gcx, 'tcx> { | |||
} | |||
} | |||
|
|||
fn need_type_info(&self, obligation: &PredicateObligation<'tcx>, ty: Ty<'tcx>) { | |||
pub fn need_type_info(&self, body_id: hir::BodyId, span: Span, ty: Ty<'tcx>) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nit for future PR: feels like this function (and its associated visitor etc) would be ripe for promotion into a sub-module error_reporting/need_type_info.rs
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sounds good Will do as soon as this lands
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Some(NodeExpr(expr)) => local_visitor.visit_expr(expr), | ||
_ => () | ||
} | ||
|
||
if let Some(pattern) = local_visitor.found_arg_pattern { | ||
err_span = pattern.span; | ||
labels.clear(); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should comment on why we are calling clear
here; maybe give before/after example
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ok
self.tables.fru_field_types.insert(node_id, ftys); | ||
} | ||
} | ||
|
||
fn visit_type_nodes(&self) { | ||
for (&id, ty) in self.fcx.ast_ty_to_ty_cache.borrow().iter() { | ||
let ty = self.resolve(ty, ResolvingTyNode(id)); | ||
let ty = self.resolve(ty, id.to_span(&self.fcx.tcx)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have a vague fear that this will add to compilation times; these spans are only needed if an error occurs. I wonder if we could change resolve
to take a closure, so that we only evaluate and get the span if an error occurs.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This was part of the reason that I suggested the Locatable
trait; you could e.g. change resolve
to take an &Locatable
(i.e., a trait object), in which case the span
field of Resolver
would change I think to the type &'cx Locatable
☔ The latest upstream changes (presumably #40570) made this pull request unmergeable. Please resolve the merge conflicts. |
Hello @nikomatsakis It took me sometime to merge Only the performance tuning remains now. Will send it tomorrow ASAP. Thanks! |
src/librustc/diagnostics.rs
Outdated
@@ -327,6 +327,54 @@ struct ListNode { | |||
This works because `Box` is a pointer, so its size is well-known. | |||
"##, | |||
|
|||
E0101: r##" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You should comment this error out.(see E0211).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done
} else { | ||
err.span_label(pattern_span, &format!("consider giving a type to pattern")); | ||
labels.push((pattern.span, format!("consider giving a type to pattern"))); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
English: consider giving the pattern a type
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm I agree this phrasing isn't the best. "consider giving a type to the pattern" or "consider giving the pattern a type" both seem fine to me.
|
||
err.span_label(cause.span, &format!("cannot infer type for `{}`", name)); | ||
let mut err_span = span; | ||
let mut labels = vec![(span, format!("cannot infer type for `{}`", name))]; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
do we ever need this message? I think a local with inference types in it is always an error at this stage. I think that if we ever find a local/argument with inference variables in it we should have E0101 with only that local spanned.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes we do. That was decided long time ago in a different issue + PR.
#38812
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Another thing that would be nice is to always use the type of the local,
E101: unable to infer the type of the local `x`
let x = vec![];
// ^ `x` can only be resolved to `Vec<_>`
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, I feel like it'd be nice to say what we do now of the type, but we never did settle on the precise format we wanted for it. (The discussion was in #39361)
@@ -66,37 +66,52 @@ impl<'a, 'gcx, 'tcx> TraitErrorKey<'tcx> { | |||
struct FindLocalByTypeVisitor<'a, 'gcx: 'a + 'tcx, 'tcx: 'a> { | |||
infcx: &'a InferCtxt<'a, 'gcx, 'tcx>, | |||
target_ty: &'a Ty<'tcx>, | |||
found_pattern: Option<&'a Pat>, | |||
hir_map: &'a hir::map::Map<'gcx>, | |||
found_local_pattern: Option<&'gcx Pat>, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why do you have separate options for found_local_pattern
and found_arg_pattern
? I'll have a single found_local
enum.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
They're not only holding target patterns, but also changing the flow of the result.
Can you share your suggestion with a concrete example please?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
enum TypeVariableHolder<'gcx> {
// later, TypeParam(...),
Argument(&'gcx Pat),
Local(&'gcx Pat),
}
struct FindLocalByTypeVisitor<'a, 'gcx: 'a + 'tcx, 'tcx: 'a> {
...,
found_variable_holder: Option<TypeVariableHolder<'gcx>>
}
And find the maximum relevant TypeVariableHolder
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It doesn't seem like it makes much difference either way to me.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually it's mainly something I want to build on.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Makes sense. I would be fine landing as is, but also fine with adopting the single enum approach -- I imagine as we add more cases, we'll probably want a single enum, as you suggest. (Though I was wondering if it may be the case that we can't detect until the end whether certain patterns are met, and hence we wouldn't be able to decide the final state until the end.) Either way, easy to evolve as needed.
Commits are squashed, and resolving logic has been optimized (relatively). ⏲ |
@@ -4146,8 +4102,6 @@ register_diagnostics! { | |||
// E0068, | |||
// E0085, | |||
// E0086, | |||
E0103, // @GuillaumeGomez: I was unable to get this error, try your best! | |||
E0104, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't think we usually remove these lints outright, but rather leave them commented out.. or at least that's what the surrounding lines seem to suggest. =)
cc @rust-lang/docs -- what is the policy when an error code is no longer issued? (And what about if a "long-form" error exists, as in the case of E0102 above?) Is this documented somewhere? (RFC 1567 doesn't seem to mention it.)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The general rule of thumb is "don't reuse error codes". It's okay to deprecate and just not output that code anymore, but reusing tends to mess with search results.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You should leave them commented out, like the previous ones.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@arielb1 commented them out.
Also instead of adding whole documentation as a comment, I've added // E0101
and // E0102
to register_diagnostics!
call.
If you have both the commented documentation and a register_diagnostics!
call for an error code, tidy
complains. Even if they're only comments.
@bors r+ |
📌 Commit 3092ac4 has been approved by |
Move E0101 and E0102 logic into new E0282 mechanism rust-lang#40013 Hello there! Previously, me and @nikomatsakis worked on error messages of uninferred locals. (rust-lang#38812) This aims to build up on that by moving certain type checks from `writeback`. With this, `E0101` and `E0102` errors are getting obsoleted and no longer thrown. They're replaced with customized versions of `E0282`s instead. ```rust error[E0282]: type annotations needed --> test.rs:2:14 | 2 | let x = |_| {}; | ^ consider giving this closure parameter a type error: aborting due to previous error ``` ```rust error[E0282]: type annotations needed --> test.rs:2:9 | 2 | let x = []; | ^ | | | consider giving `x` a type | cannot infer type for `[_; 0]` error: aborting due to previous error ``` - I think we need to change our way of type name resolving in relevant places, because that `[_; 0]` looks horrible IMHO. - I'm not terribly happy with the note ordering of errors. So please do point to code that might help me accomplish this. Tests of `E0101` and `E0102` are getting converted from `compile-fail` to `ui` tests. Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones. Please do comment on messages, layouts and other details. Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻
Move E0101 and E0102 logic into new E0282 mechanism rust-lang#40013 Hello there! Previously, me and @nikomatsakis worked on error messages of uninferred locals. (rust-lang#38812) This aims to build up on that by moving certain type checks from `writeback`. With this, `E0101` and `E0102` errors are getting obsoleted and no longer thrown. They're replaced with customized versions of `E0282`s instead. ```rust error[E0282]: type annotations needed --> test.rs:2:14 | 2 | let x = |_| {}; | ^ consider giving this closure parameter a type error: aborting due to previous error ``` ```rust error[E0282]: type annotations needed --> test.rs:2:9 | 2 | let x = []; | ^ | | | consider giving `x` a type | cannot infer type for `[_; 0]` error: aborting due to previous error ``` - I think we need to change our way of type name resolving in relevant places, because that `[_; 0]` looks horrible IMHO. - I'm not terribly happy with the note ordering of errors. So please do point to code that might help me accomplish this. Tests of `E0101` and `E0102` are getting converted from `compile-fail` to `ui` tests. Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones. Please do comment on messages, layouts and other details. Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻
Move E0101 and E0102 logic into new E0282 mechanism rust-lang#40013 Hello there! Previously, me and @nikomatsakis worked on error messages of uninferred locals. (rust-lang#38812) This aims to build up on that by moving certain type checks from `writeback`. With this, `E0101` and `E0102` errors are getting obsoleted and no longer thrown. They're replaced with customized versions of `E0282`s instead. ```rust error[E0282]: type annotations needed --> test.rs:2:14 | 2 | let x = |_| {}; | ^ consider giving this closure parameter a type error: aborting due to previous error ``` ```rust error[E0282]: type annotations needed --> test.rs:2:9 | 2 | let x = []; | ^ | | | consider giving `x` a type | cannot infer type for `[_; 0]` error: aborting due to previous error ``` - I think we need to change our way of type name resolving in relevant places, because that `[_; 0]` looks horrible IMHO. - I'm not terribly happy with the note ordering of errors. So please do point to code that might help me accomplish this. Tests of `E0101` and `E0102` are getting converted from `compile-fail` to `ui` tests. Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones. Please do comment on messages, layouts and other details. Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻
Move E0101 and E0102 logic into new E0282 mechanism rust-lang#40013 Hello there! Previously, me and @nikomatsakis worked on error messages of uninferred locals. (rust-lang#38812) This aims to build up on that by moving certain type checks from `writeback`. With this, `E0101` and `E0102` errors are getting obsoleted and no longer thrown. They're replaced with customized versions of `E0282`s instead. ```rust error[E0282]: type annotations needed --> test.rs:2:14 | 2 | let x = |_| {}; | ^ consider giving this closure parameter a type error: aborting due to previous error ``` ```rust error[E0282]: type annotations needed --> test.rs:2:9 | 2 | let x = []; | ^ | | | consider giving `x` a type | cannot infer type for `[_; 0]` error: aborting due to previous error ``` - I think we need to change our way of type name resolving in relevant places, because that `[_; 0]` looks horrible IMHO. - I'm not terribly happy with the note ordering of errors. So please do point to code that might help me accomplish this. Tests of `E0101` and `E0102` are getting converted from `compile-fail` to `ui` tests. Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones. Please do comment on messages, layouts and other details. Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻
⌛ Testing commit 3092ac4 with merge e49664e... |
@bors retry prioritizing rollup which includes this |
Move E0101 and E0102 logic into new E0282 mechanism #40013 Hello there! ## What's this? Previously, me and @nikomatsakis worked on error messages of uninferred locals. (#38812) This aims to build up on that by moving certain type checks from `writeback`. With this, `E0101` and `E0102` errors are getting obsoleted and no longer thrown. They're replaced with customized versions of `E0282`s instead. ## Sample Error Messages #### `E0101` is getting converted into: ```rust error[E0282]: type annotations needed --> test.rs:2:14 | 2 | let x = |_| {}; | ^ consider giving this closure parameter a type error: aborting due to previous error ``` #### `E0102` is getting converted into: ```rust error[E0282]: type annotations needed --> test.rs:2:9 | 2 | let x = []; | ^ | | | consider giving `x` a type | cannot infer type for `[_; 0]` error: aborting due to previous error ``` ## Annoyances - I think we need to change our way of type name resolving in relevant places, because that `[_; 0]` looks horrible IMHO. - I'm not terribly happy with the note ordering of errors. So please do point to code that might help me accomplish this. ## Tests Tests of `E0101` and `E0102` are getting converted from `compile-fail` to `ui` tests. ## Documentation Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones. Please do comment on messages, layouts and other details. ## Appreciation Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻
☀️ Test successful - status-appveyor, status-travis |
Hello there!
What's this?
Previously, me and @nikomatsakis worked on error messages of uninferred locals. (#38812)
This aims to build up on that by moving certain type checks from
writeback
.With this,
E0101
andE0102
errors are getting obsoleted and no longer thrown.They're replaced with customized versions of
E0282
s instead.Sample Error Messages
E0101
is getting converted into:E0102
is getting converted into:Annoyances
[_; 0]
looks horrible IMHO.Tests
Tests of
E0101
andE0102
are getting converted fromcompile-fail
toui
tests.Documentation
Please help me with documentation update. There are some confusing places that needed an update but I'm not sure if I did the right ones.
Please do comment on messages, layouts and other details.
Appreciation
Huge thanks goes to @nikomatsakis for being a patient and humble mentor along this long journey. 🍻