-
Notifications
You must be signed in to change notification settings - Fork 13.5k
Tweak some diagnostic spans #59084
New issue
Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? # to your account
Tweak some diagnostic spans #59084
Conversation
r? @cramertj (rust_highfive has picked a reviewer for you, use r? to override) |
r? @davidtwco |
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.
LGTM, one minor question, feel free to r=me.
@@ -5115,7 +5115,7 @@ impl<'a> Resolver<'a> { | |||
// extra for the comma. | |||
span.lo().0 - (prev_comma.as_bytes().len() as u32) - 1 | |||
)); | |||
err.span_suggestion( | |||
err.tool_only_span_suggestion( |
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'm unsure about this change. I agree that it makes the error much cleaner, but without the help: ...
message how would someone know that they can automatically have the problem fixed?
(as a slight aside, I think we should try use try cc groups like @rust-lang/wg-diagnostics
when landing PRs like this or the one that introduced tool_only_span_suggestion
- I had no idea this existed and it'd have been nice to follow along with 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.
tool_only_span_diagnostics
was landed over the all hands because it was needed for a different change, so it's pretty recent.
Tools still will see the same json output, so vscode and rustfix would still be able to apply these. I guess I hadn't thought of the use case of someone running rustfix after running rustc iteratively. Should we mark these some other way? Cyan asterisk next to the error or something?
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.
My only concern is that users who know when they see help: ...
that it is something that can be automatically applied won't know that for this error.
Something like a cyan asterisk would be ideal to draw attention to that (it might even raise awareness of rustfix being able to fix these things if everyone's error messages had a cyan asterisk suddenly appear). I wouldn't block this PR on that, so I'll r+ this and we can file a follow-up.
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.
Filed #59103.
@bors r+ |
📌 Commit 63432949565d73020222ba99fb487f855b4b0565 has been approved by |
@davidtwco added a small tweak to fix some less than stellar spans and marked the other duplicate import suggestion as tool only. @bors r=davidtwco |
📌 Commit a00fa50a3a47274416a3a1d404eb3a3a3afaba07 has been approved by |
This comment has been minimized.
This comment has been minimized.
a00fa50
to
f1ee5bb
Compare
@bors r=davidtwco rebased |
📌 Commit f1ee5bb2c5029a452c7ae8f465f87e3278a640c9 has been approved by |
☔ The latest upstream changes (presumably #58899) made this pull request unmergeable. Please resolve the merge conflicts. |
f1ee5bb
to
b53ca90
Compare
This comment has been minimized.
This comment has been minimized.
@bors r=davidtwco |
📌 Commit d1656f1 has been approved by |
Tweak some diagnostic spans
Tweak some diagnostic spans
☔ The latest upstream changes (presumably #58929) made this pull request unmergeable. Please resolve the merge conflicts. |
d1656f1
to
59f0f2e
Compare
@bors r=davidtwco |
📌 Commit 59f0f2e has been approved by |
☀️ Test successful - checks-travis, status-appveyor |
No description provided.