Closed
Conversation
This is continuation on the "copy suggestions". The logic is as follows:
- If the type is already `Clone` - clone it
- If it can be `Copy` - make it copy
- If it can be `Clone` - make it clone and the clone it
- Otherwise feel sad
Note that clone suggestions may be incorrect in many cases, like for
example `S{t}; t` that suggests `S{t.clone()}; t`.
See `use_of_moved_value_clone_suggestions_bad.rs` for more problematic
cases.
rustbot
added
T-compiler
rust-highfive
added
the
S-waiting-on-review
WaffleLapkin
commented
Mar 19, 2022
Comment on lines
+41
to
+48
Member
Author
There was a problem hiding this comment.
To be clear: these suggestions could be merged together, it just requires some work (we need the suggest_constraining_type_params to allow adding to the same suggestion/changing the text/etc).
This comment has been minimized.
This comment has been minimized.
WaffleLapkin
force-pushed
the
clone-suggestions
branch
from
4084e1b to
4c5de71
Compare
Collaborator
|
The job Click to see the possible cause of the failure (guessed by this bot) |
estebank
reviewed
Mar 22, 2022
Comment on lines
+11
to
+14
| help: consider cloning `x.f` | ||
| | | ||
| LL | Foo {f.clone()} => {} | ||
| | ++++++++ |
Contributor
There was a problem hiding this comment.
This seems to be an incorrect suggestion. We'll need to check that we're not in a pattern, we can only call function in expressions.
Member
Author
There was a problem hiding this comment.
Yes, but I couldn't figure a way to check :')
estebank
reviewed
Mar 22, 2022
| }) | ||
| .collect() | ||
| }); | ||
| let copy_did = tcx.lang_items().copy_trait().unwrap(); |
Contributor
There was a problem hiding this comment.
There's a method, I think called expect_lang_item or similar (don't have the editor open right now to find it) that does that unwrap for you in a non-ICEy way.
Collaborator
|
☔ The latest upstream changes (presumably #95257) made this pull request unmergeable. Please resolve the merge conflicts. |
Contributor
|
I still need to look at how we can avoid suggesting this in patterns. |
Contributor
|
I would like to see this land, but don't have enough knowledge of the mir to help here. @JakobDegen I see that you have worked in |
Contributor
|
As far as I know, MIR currently does not contain enough information to make this distinction. I took a look at the let F { y } = x;The span of the relevant statement is still the pattern and we get the following relevant MIR FakeRead(ForLet(None), _1); // scope 0 at test.rs:6:9: 6:10
StorageDead(_4); // scope 0 at test.rs:6:38: 6:39
StorageLive(_5); // scope 1 at test.rs:8:15: 8:16
_5 = move (_1.0: std::string::String); // scope 1 at test.rs:8:15: 8:16while this version let y = x.f;has the span of the statement on the expression and produces: FakeRead(ForLet(None), _1); // scope 0 at test.rs:6:9: 6:10
StorageDead(_4); // scope 0 at test.rs:6:38: 6:39
StorageLive(_5); // scope 1 at test.rs:8:9: 8:10
_5 = move (_1.0: std::string::String); // scope 1 at test.rs:8:13: 8:16
FakeRead(ForLet(None), _5); // scope 1 at test.rs:8:9: 8:10The There are a number of places in MIR where we already store additional info for the purpose of diagnostics. Possibly this could be done for this scenario too, but I don't quite see a nice way of doing it. In any case, I am also somewhat out of my depth here. cc @oli-obk who might know how to go about this. |
Contributor
|
I don't know of a good solution, but you can try dumping the |
Contributor
|
I thought about this more, and adding things to the local decls might be a bit of a challenge. The problem is destructuring assignment: struct Foo {
f: i32,
}
fn touch() {}
pub fn f() {
let x = Foo { f: 10 };
let mut y = 5;
touch(); // Just want to separate basic blocks
y = x.f;
touch();
Foo { f: y } = x;
touch();
}For the regular assignment: bb1: {
StorageDead(_3); // scope 2 at test.rs:11:12: 11:13
StorageLive(_4); // scope 2 at test.rs:13:9: 13:12
_4 = (_1.0: i32); // scope 2 at test.rs:13:9: 13:12
_2 = move _4; // scope 2 at test.rs:13:5: 13:12
StorageDead(_4); // scope 2 at test.rs:13:11: 13:12
StorageLive(_5); // scope 2 at test.rs:15:5: 15:12
_5 = touch() -> [return: bb2, unwind: bb4]; // scope 2 at test.rs:15:5: 15:12
// mir::Constant
// + span: test.rs:15:5: 15:10
// + literal: Const { ty: fn() {touch}, val: Value(Scalar(<ZST>)) }
}And for the destructuring assignment: bb2: {
StorageDead(_5); // scope 2 at test.rs:15:12: 15:13
StorageLive(_6); // scope 2 at test.rs:17:5: 17:21
StorageLive(_7); // scope 2 at test.rs:17:14: 17:15
_7 = (_1.0: i32); // scope 2 at test.rs:17:14: 17:15
StorageLive(_8); // scope 3 at test.rs:17:14: 17:15
_8 = _7; // scope 3 at test.rs:17:14: 17:15
_2 = move _8; // scope 3 at test.rs:17:14: 17:15
StorageDead(_8); // scope 3 at test.rs:17:14: 17:15
_6 = const (); // scope 2 at test.rs:17:5: 17:21
StorageDead(_7); // scope 2 at test.rs:17:20: 17:21
StorageDead(_6); // scope 2 at test.rs:17:21: 17:22
StorageLive(_9); // scope 2 at test.rs:19:5: 19:12
_9 = touch() -> [return: bb3, unwind: bb4]; // scope 2 at test.rs:19:5: 19:12
// mir::Constant
// + span: test.rs:19:5: 19:10
// + literal: Const { ty: fn() {touch}, val: Value(Scalar(<ZST>)) }
}The only real difference is the extra temporary, and that doesn't seem reliable. Possibly we could add some information to the local decl of the temporary though? |
Contributor
Can we prototype that and see if that would affect perf in any way? |
Member
|
Triage: @rustbot author |
rustbot
added
S-waiting-on-author
Member
|
@WaffleLapkin @rustbot label: +S-inactive |
rustbot
added
the
S-inactive
Contributor
|
For what is worth, I really want to see this land eventually, but for that we need to figure out how to avoid the invalid suggestions when involving patterns. |
Member
Author
|
@compiler-errors proposed a possible solution recently, I'm currently investigating if it works 👀 |
Member
Author
|
For posterity sake: this seems to have been implemented in #103908 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This continues the work from #94375 by also suggestion cloning sometimes, instead of coping. The rules are as follows:
Clone, suggest adding.clone()callsCopy, suggest adding bounds requited to make itCopyClone, suggest adding bounds requited to make itCloneand suggest adding.clone()callsFor example for this code:
we now produce the following diagnostics:
Which is quite nice in my opinion!
However, there are some corner-cases where the suggestion is wrong and I don't know how to fix it. See
use_of_moved_value_clone_suggestions_bad.rstest for examples of broken suggestions that I could find.r? @estebank
@rustbot label +A-diagnostics +A-suggestion-diagnostics +C-enhancement