Emit a warning instead of an error for unknown rule selectors by ntBre · Pull Request #26113 · astral-sh/ruff

ntBre · 2026-06-17T19:03:13Z

Summary

This PR fixes #14443 by deferring rule parsing. This allows deserializing configuration files and
CLI arguments with unknown rule selectors and enables us to emit a warning instead of a hard error
when encountering an unknown selector. This will also help with #25887 because we can also defer
parsing until we have an accurate preview setting for checking whether human-readable names should
be allowed. I've left some of this infrastructure in place in this PR, with the preview argument
currently unused.

Test Plan

Existing tests updated to show warnings instead of errors

Summary -- This PR fixes #14443 by deferring rule parsing. This allows deserializing configuration files and CLI arguments with unknown rule selectors and enables us to emit a warning instead of a hard error when encountering an unknown selector. This will also help with #25887 because we can also defer parsing until we have an accurate `preview` setting for checking whether human-readable names should be allowed. I've left some of this infrastructure in place in this PR, with the `preview` argument currently unused. Test Plan -- Existing tests updated to show warnings instead of errors

ntBre · 2026-06-17T19:09:01Z

-                // More specified selectors take precedence over less specified selectors
-                .collect()
-        };
+        let mut safety_override_map: FxHashMap<Rule, (Specificity, Override)> =


This is probably the least mechanical change in the PR. My initial naive approach was just calling selector.resolve in the existing Specificity::iter().flat_map but this would resolve/parse each selector once per Specificity (of which there are 8 variants), so this instead takes one loop over the selectors and stores their specificity in the map to avoid multiple resolutions.

ntBre · 2026-06-17T19:14:38Z

-        }
-        if !unknown.is_empty() {
-            *contains_invalid_settings = true;
-            tracing::error!("Unknown rule selectors found in `{key}`: {unknown:?}");


It's unfortunate to lose this more specific error message, but I think we'll recover it with something like ty's RangedValue in the future.

Did you explore what it would take to preserve this information. Ideally, we'd keep source (which file, cli) + key

astral-sh-bot · 2026-06-17T19:23:57Z

`ruff-ecosystem` results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

✅ ecosystem check detected no linter changes.

Formatter (stable)

✅ ecosystem check detected no format changes.

Formatter (preview)

✅ ecosystem check detected no format changes.

MichaReiser

Nice, thank you.

Unfortunately, the way this is implemented now is a breaking change. Users might rely on Ruff erroring for unknown rule codes. I think we have to preserve that behavior at least in non preview mode for now.

Did you explore trying to preserve the source? Not the exact location in the TOML, but at least enough information so that we can point the user to the right configuration file and maybe key.

MichaReiser · 2026-06-18T06:27:11Z

+    for args in [
+        ["--select", "F481"],
+        ["--extend-select", "F481"],
+        ["--ignore", "F481"],
+        ["--fixable", "F481"],
+        ["--extend-fixable", "F481"],
+        ["--unfixable", "F481"],
+        ["--config", "lint.ignore=['F481']"],
+        ["--config", "lint.extend-safe-fixes=['F481']"],
+        ["--config", "lint.extend-unsafe-fixes=['F481']"],
+        ["--per-file-ignores", "test.py:F481"],
+        ["--extend-per-file-ignores", "test.py:F481"],
+        ["--config", "lint.per-file-ignores={'test.py'=['F481']}"],
+        [
+            "--config",
+            "lint.extend-per-file-ignores={'test.py'=['F481']}",
+        ],
+    ] {
+        let output = fixture
+            .check_command()
+            .args(["--select", "F401"])
+            .args(args)
+            .arg("test.py")
+            .output()?;
+        assert_eq!(output.status.code(), Some(1), "arguments: {args:?}");
+        assert!(
+            str::from_utf8(&output.stdout)?.contains("F401"),
+            "arguments: {args:?}"
+        );
+        assert_eq!(
+            str::from_utf8(&output.stderr)?,
+            "warning: Invalid rule selector: `F481`\n",
+            "arguments: {args:?}"
+        );
+    }
+


Nit: I'd write a test helper here that returns str::from_utf8(output.stderr) for a given arg and value instead of using a for loop in the test. For loops over test cases is very annoying in tests because it isn't obvious which specific case failed. The test helper avoids the main boilerplate, but you can still have a separate test function for each case and have the assert_eq in those test functions.

Gotcha, that makes sense. I was initially put off by this too when Codex wrote it but figured it was okay in this instance. I can probably just turn them into test_cases instead of a loop, since we're currently asserting on the code, stdout, and stderr. Or we could just snapshot them I guess.

I personally dislike test cases because I can’t run those tests from within my ide. But I wouldn’t object

I will split them up. I need to customize the test names anyway, and I think I'd prefer just setting the function name over customizing that through test_case too.

MichaReiser · 2026-06-18T06:28:16Z

+}
+
 #[test]
 fn valid_toml_but_nonexistent_option_provided_via_config_argument() {


We may need to change this test. I'm not sure it still captures what it originally intended.

MichaReiser · 2026-06-18T06:28:46Z

-    error: invalid value 'PREVIEW' for '--select <RULE_CODE>'
-
-    For more information, try '--help'.
+    warning: Invalid rule selector: `PREVIEW`


Nit: Do we know whether the rule was removed? If so, could we emit a more specific error message?

We know if Rules are removed from their group method, but I don't think we keep track of these removed selectors. I've added an error kind for removed selectors, but it currently just checks against the hard-coded list of PREVIEW and NURSERY.

MichaReiser · 2026-06-18T06:33:33Z

+                // Rules in preview are included here even if preview mode is disabled; it's safe to
+                // ignore disabled rules.


Can you expand on what you mean by safe here? Do we not need to warn about unknown rules in per_file_ignores? What makes it safe?

This was a pre-existing comment that was previously in PerFileIgnore::new. It means that it's "safe" to use all_rules here even though it doesn't filter out preview rules because preview rules are already ignored when preview itself is disabled.

Maybe the comment can be deleted anyway, but I added it when originally working on per-file-ignores I think.

MichaReiser · 2026-06-18T06:35:07Z

+}
+
+impl UnresolvedRuleSelector {
+    pub fn resolve(&self, _preview: PreviewMode) -> Option<RuleSelector> {


Why is _preview unused here? Remove?

I'm planning to use this immediately in #25887, so I wanted to make sure I had easy access to a PreviewMode wherever this was used. I can still remove it for now if you want.

MichaReiser · 2026-06-18T06:36:30Z

+    pub fn resolve(&self, _preview: PreviewMode) -> Option<RuleSelector> {
+        RuleSelector::from_str(&self.selector)
+            .inspect_err(|_| {
+                warn_user_once_by_message!("Invalid rule selector: `{}`", self.selector);


Maybe Unknown rule selector

It would also be nice if we could tell users where this rule selector was specified. Imagine a large mono repository with a 1000 configs. How are you supposed to find the rule selector (okay, you can do a grep, but this only works if all projects are using the same Ruff version)

MichaReiser · 2026-06-18T06:39:28Z

-        }
-        if !unknown.is_empty() {
-            *contains_invalid_settings = true;
-            tracing::error!("Unknown rule selectors found in `{key}`: {unknown:?}");


Did you explore what it would take to preserve this information. Ideally, we'd keep source (which file, cli) + key

MichaReiser · 2026-06-18T06:46:04Z

+            );
+
+        for (selector, o) in selectors {
+            let Some(selector) = selector.resolve(preview_options.mode) else {


IMO, doing the logging in resolve feels a bit implicit. Maybe resolve_and_warn or move the error handling out of resolve (I don't know how annoying that would be).

Or, what we do in ty. We have a custom error, and the error has helper methods. E.g. this could be

let selector = match selector.resolve(preview_options.mode) { Ok(selector) => selector, Err(err) => { err.log_warning("unsafe_fixes"); continue; } };

Ah yeah, this doesn't seem as bad as I thought. This is only called in 3 places, so I'll just move the logging out of resolve.

ntBre · 2026-06-18T13:45:57Z

Did you explore trying to preserve the source? Not the exact location in the TOML, but at least enough information so that we can point the user to the right configuration file and maybe key.

Not really beyond reading how ty does it. I tried to structure things so that this would be easy in the future but thought we might leave that for a follow-up. Maybe it makes more sense to do it here, though, in line with a couple of your inline comments.

ntBre · 2026-06-18T22:51:42Z

I adapted a lot of ty's RangedValue source information and addressed most of the other comments, but I'm still working through some of Codex's review findings. Namely, it's now possible to panic here in the editor if invalid settings are present:

ruff/crates/ruff_server/src/session/index/ruff_settings.rs

Lines 117 to 119 in a26853c

    
           Self::with_editor_settings(editor_settings, root, configuration).expect( 
        
               "editor configuration should merge successfully with default configuration", 
        
           )

because the resolution is actually deferred too long for the current structure. It's looking kind of involved to work around this, so I still need to think about it a bit more.

I think the new snapshot file names are also invalid on Windows, but that's obviously easier to fix.

MichaReiser · 2026-06-19T05:47:46Z

This sounds very much like #25024. I think the fix there is easy, see my comment.

astral-sh-bot · 2026-06-24T15:50:43Z

Typing conformance results

No changes detected ✅

Current numbers

The percentage of diagnostics emitted that were expected errors held steady at 94.47%. The percentage of expected errors that received a diagnostic held steady at 89.10%. The number of fully passing files held steady at 95/134.

astral-sh-bot · 2026-06-24T15:52:07Z

Memory usage report

Memory usage unchanged ✅

astral-sh-bot · 2026-06-24T15:52:52Z

`ecosystem-analyzer` results

Lint rule	Added	Removed
`unresolved-reference`	0	13
`invalid-return-type`	3	0
`invalid-argument-type`	1	0
`unresolved-attribute`	1	0
Total	5	13

Raw diff (18 changes)

hydpy (https://github.com/hydpy-dev/hydpy)
- hydpy/core/parametertools.py:3050:23 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/sequencetools.py:2123:23 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/sequencetools.py:3392:23 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/sequencetools.py:3663:23 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/sequencetools.py:3802:22 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/sequencetools.py:4016:27 error[unresolved-reference] Name `__class__` used when not defined
- hydpy/core/variabletools.py:2292:27 error[unresolved-reference] Name `__class__` used when not defined
+ hydpy/core/variabletools.py:2293:13 error[unresolved-attribute] Object of type `<super: <class 'MixinFixedShape'>, type[Self@shape]>` has no attribute `shape`
- hydpy/models/evap/evap_logs.py:77:22 error[unresolved-reference] Name `__class__` used when not defined
+ hydpy/models/evap/evap_logs.py:78:25 error[invalid-argument-type] Argument to function `Variable.shape` is incorrect: Expected `int | tuple[int, ...]`, found `tuple[Literal[1], int | tuple[int, ...]]`

pywin32 (https://github.com/mhammond/pywin32)
- pythonwin/pywin/test/test_pywin.py:141:31 error[unresolved-reference] Name `__class__` used when not defined
- pythonwin/pywin/test/test_pywin.py:141:55 error[unresolved-reference] Name `__class__` used when not defined
- pythonwin/pywin/test/test_pywin.py:287:25 error[unresolved-reference] Name `__class__` used when not defined

rotki (https://github.com/rotki/rotki)
+ rotkehlchen/tasks/historical_balances.py:186:20 error[invalid-return-type] Return type does not match returned value: expected `list[tuple[Bucket, Literal[EventDirection.IN, EventDirection.OUT]]]`, found `list[tuple[Bucket, Literal[EventDirection.IN, EventDirection.OUT]] | tuple[Self@from_event, EventDirection]]`
+ rotkehlchen/tasks/historical_balances.py:195:16 error[invalid-return-type] Return type does not match returned value: expected `list[tuple[Bucket, Literal[EventDirection.IN, EventDirection.OUT]]]`, found `list[tuple[Bucket, Literal[EventDirection.IN, EventDirection.OUT]] | tuple[Self@from_event, EventDirection]]`

scrapy (https://github.com/scrapy/scrapy)
- tests/test_item.py:293:28 error[unresolved-reference] Name `__class__` used when not defined

steam.py (https://github.com/Gobot1234/steam.py)
+ steam/ext/commands/cooldown.py:52:43 error[invalid-return-type] Function can implicitly return `None`, which is not assignable to return type `BucketTypeType`

sympy (https://github.com/sympy/sympy)
- sympy/core/cache.py:209:24 error[unresolved-reference] Name `__class__` used when not defined

Full report with detailed diff (timing results)

## Summary Record `ruff_ranged_value` as configured for crates.io trusted publishing after #26113 made it publishable. The bootstrap script uses `.known-crates` as a persistent checkpoint and skips entries on later runs. Checking in the generated result makes subsequent bootstrap runs a no-op until another workspace package becomes publishable. A dry run confirms that all 37 publishable workspace packages are recorded and no setup work remains.

## Summary This PR enables using rule names as selectors in preview by also attempting to parse a selector as a human-readable name in `UnresolvedRuleSelector::resolve`. If this succeeds and preview is enabled, the rule is activated, and if preview is disabled you get a custom error message. ## Test Plan A few new CLI tests. I figured the tests from #26113 cover the shared behavior for the other selectors well enough and just focused on `select` as an example for both the CLI and config file.

ntBre added 2 commits June 17, 2026 14:44

copy the test from brent/human-readable-selectors

eb4a0eb

ntBre added the rule-selection Related to enabling or disabling rules label Jun 17, 2026

ntBre commented Jun 17, 2026

View reviewed changes

ntBre marked this pull request as ready for review June 17, 2026 19:40

ntBre requested a review from MichaReiser June 17, 2026 19:40

MichaReiser requested changes Jun 18, 2026

View reviewed changes

o -> safety_override

3fbb9b8

skip explicit unsafe check

b489608

ntBre marked this pull request as draft June 18, 2026 13:46

ntBre added the preview Related to preview mode features label Jun 18, 2026

ntBre added 14 commits June 18, 2026 15:12

test table -> test_case

3dc170c

update valid_toml_but_nonexistent_option_provided_via_config_argument

3ce65ef

switch to snapshots

e6bc749

preview-gate warning behavior

338f775

remove unused deserialize impl for PatternPrefixPair

7e4747a

copy over ValueSourceGuard

5ebc2b5

avoid missing span errors for DeprecatedTopLevelLintOptions

6e58f65

add cli constructor, use in tests

19dcadb

use source and setting to improve error message

a2cc9c0

don't derive Eq and Hash

da85031

add range getter

dcc2d65

ValueSourceGuard -> RuleSelectorSourceGuard

3da50d8

clean up unused snapshots

7a87afe

add preview cases too

a26853c

ntBre added 7 commits June 24, 2026 10:21

ruff_ranged_value crate

b8cbb44

use RangedValue in ruff

0fc7305

shear

bbf5576

fix optional dep

070d7cd

revert Combined for RangedValue dependency

7aa7590

docs

fe3d37a

generate readme

420c427

ntBre requested review from a team as code owners June 24, 2026 15:48

tidy some imports

6f47201

ntBre removed request for a team June 24, 2026 16:03

ntBre merged commit 62859df into main Jun 24, 2026
63 checks passed

ntBre deleted the brent/defer-rule-parsing branch June 24, 2026 16:13

charliermarsh mentioned this pull request Jun 24, 2026

Record configured ruff_ranged_value package #26333

Merged

ntBre mentioned this pull request Jun 24, 2026

Allow human-readable names in rule selectors #25887

Merged

BrewTestBot mentioned this pull request Jun 25, 2026

ruff 0.15.20 Homebrew/homebrew-core#289873

Merged

		// Rules in preview are included here even if preview mode is disabled; it's safe to
		// ignore disabled rules.

Uh oh!

Conversation

ntBre commented Jun 17, 2026

Summary

Test Plan

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

astral-sh-bot Bot commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ruff-ecosystem results

Linter (stable)

Linter (preview)

Formatter (stable)

Formatter (preview)

Uh oh!

MichaReiser left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ntBre commented Jun 18, 2026

Uh oh!

ntBre commented Jun 18, 2026

Uh oh!

MichaReiser commented Jun 19, 2026

Uh oh!

astral-sh-bot Bot commented Jun 24, 2026

No changes detected ✅

Uh oh!

astral-sh-bot Bot commented Jun 24, 2026

Memory usage report

Uh oh!

astral-sh-bot Bot commented Jun 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ecosystem-analyzer results

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

astral-sh-bot Bot commented Jun 17, 2026 •

edited

Loading

`ruff-ecosystem` results

astral-sh-bot Bot commented Jun 24, 2026 •

edited

Loading

`ecosystem-analyzer` results