Repository navigation
Conversation
|
These commits modify the If this was unintentional then you should revert the changes before this PR is merged. |
This comment has been minimized.
This comment has been minimized.
bd12853 to
791e562
Compare
|
The parser was modified, potentially altering the grammar of (stable) Rust cc @fmease |
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
I'm leaning towards r+, but am slightly concerned by this new flag option being insta-stable. Could we add a check for -Zunstable-options being present before allowing the output? It could even still allow changing the rendering (even on stable), as long as an error is always emitted complaining that the feature is unstable. That way people can already start using it, while giving no stability guarantees because having the flag always enabled would cause it to never permit a compile to succeed.
| @@ -33,6 +33,7 @@ use crate::{ | |||
| /// Describes the way the content of the `rendered` field of the json output is generated | |||
| #[derive(Clone, Copy, Debug, PartialEq, Eq)] | |||
| pub struct HumanReadableErrorType { | |||
There was a problem hiding this comment.
Non-blocker. We should be moving towards an enum, I think.
There was a problem hiding this comment.
I agree that we should move to an enum, and I tried to do so, but I ran into issues discerning what should be variants vs. options e.g., is short its own rendering format (variant), or is it an option on graphics/no-graphics?
There was a problem hiding this comment.
I think that (at least eventually) we likely want to fully enumerate all of the cases (so short would be its own thing). The only one that is orthogonal is color, IMO. To be clear, I don't mind not doing that in this PR!
| //! Check that compile errors are formatted in the "no-graphics" style | ||
| //! when `--error-format=human-no-graphics` is used. | ||
| //@ compile-flags: --error-format=human-no-graphics |
There was a problem hiding this comment.
Is there no gate for -Znightly-features for --error-format=human-no-graphics, or for that matter the previously existing --error-format=human-unicode? I thought we hadn't stabilized that last one yet (but I can see on stable rustc that it works).
The only reason I am concerned about it is that I might want to have a round of discussion on what the flag name should be long term. I am personally biased towards this new mode being called --error-format=text-only or --error-format=no-graphics.
There was a problem hiding this comment.
Is there no gate for
-Znightly-featuresfor--error-format=human-no-graphics, or for that matter the previously existing--error-format=human-unicode?
-Zunstable-options isn't required in UI tests since RUSTC_BOOTSTRAP=1 is set for all UI tests. The docs I linked to say that act-as-stable can be used in UI tests to verify stable-only behavior; I can add a test that uses it if you want.
I thought we hadn't stabilized that last one yet (but I can see on stable rustc that it works).
Maybe I did something wrong, but using --error-format=human-unicode on +stable gives me an error, but not on nightly:
❯ echo 'fn main(){let x;}' | rustc +stable - --error-format=human-unicode
error: `--error-format=human-unicode` is unstable❯ echo 'fn main(){let x;}' | rustc +nightly - --error-format=human-unicode
error[E0282]: type annotations needed
╭▸ <anon>:1:15
│
1 │ fn main(){let x;}
│ ━
╰╴
help: consider giving `x` an explicit type
╭╴
1 │ fn main(){let x: /* Type */;}
╰╴ ++++++++++++
error: aborting due to 1 previous errorThe only reason I am concerned about it is that I might want to have a round of discussion on what the flag name should be long term. I am personally biased towards this new mode being called
--error-format=text-onlyor--error-format=no-graphics.
I personally prefer --error-format=no-graphics and would be happy to switch to it if it seems like the right thing to do. I only went with --error-format=human-no-graphics to match the existing naming convention(s).
There was a problem hiding this comment.
Can we add two more svg tests showing the interaction of color and no-color with the new diagnostic render?
There was a problem hiding this comment.
I can add an SVG test for color, but no-color always uses .stderr, as far as I can tell. The only way I could find to get no-color to use an SVG was to also enable -Zterminal-urls=yes.
There was a problem hiding this comment.
I added a test that shows color (svg), no-color (stderr), and terminal-urls (SVG, with color). Let me know if they are what you were looking for
791e562 to
c34bdf9
Compare
|
This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed. Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers. |
|
@bors r+ |
…tebank Add support for `annotate-snippets`'s `no-graphics` output format Tracking Issue: rust-lang#165164 MCP: rust-lang/compiler-team#1045 This PR adds the initial support for `annotate-snippets`'s [`no-graphics` output format](https://docs.rs/annotate-snippets/latest/annotate_snippets/struct.Renderer.html#method.no_graphics). The `no-graphics` output format doesn't show source snippets or use any graphical decorations and is primarily designed for Text-To-Speech and Braille displays. The goal at this time is to get `no-graphics` in front of users so we can iterate on and improve the output. --- 🤖 LLM involvement: I used an LLM to review the commits before submitting r? @estebank
…tebank Add support for `annotate-snippets`'s `no-graphics` output format Tracking Issue: rust-lang#165164 MCP: rust-lang/compiler-team#1045 This PR adds the initial support for `annotate-snippets`'s [`no-graphics` output format](https://docs.rs/annotate-snippets/latest/annotate_snippets/struct.Renderer.html#method.no_graphics). The `no-graphics` output format doesn't show source snippets or use any graphical decorations and is primarily designed for Text-To-Speech and Braille displays. The goal at this time is to get `no-graphics` in front of users so we can iterate on and improve the output. --- 🤖 LLM involvement: I used an LLM to review the commits before submitting r? @estebank
|
💔 I suspect this PR failed tests as part of a rollup After fixing the problem, consider running a try job for the failed job before re-approving. Link to failure: #165742 (comment) Probably this PR? |
|
This pull request was unapproved. This PR was contained in a rollup (#165742), which was unapproved. |
Add support for `annotate-snippets`'s `no-graphics` output format try-job: test-x86_64-gnu-aux
This comment has been minimized.
This comment has been minimized.
|
💔 Test for 5220576 failed: CI. Failed job:
|
|
The job Click to see the possible cause of the failure (guessed by this bot) |
Tracking Issue: #165164
MCP: rust-lang/compiler-team#1045
This PR adds the initial support for
annotate-snippets'sno-graphicsoutput format. Theno-graphicsoutput format doesn't show source snippets or use any graphical decorations and is primarily designed for Text-To-Speech and Braille displays. The goal at this time is to getno-graphicsin front of users so we can iterate on and improve the output.🤖 LLM involvement: I used an LLM to review the commits before submitting
r? @estebank