v0.4.5

Changes

Ruff's language server is now in Beta

v0.4.5 marks the official Beta release of ruff server, an integrated language server built into Ruff. ruff server supports the same feature set as ruff-lsp, powering linting, formatting, and code fixes in Ruff's editor integrations -- but with superior performance and no installation required. We'd love your feedback!

You can enable ruff server in the VS Code extension today.

To read more about this exciting milestone, check out our blog post!

Rule changes

• [flake8-future-annotations] Reword future-rewritable-type-annotation (FA100) message (#11381)
• [isort] Expanded the set of standard-library modules to include _string, etc. (#11374)
• [pycodestyle] Consider soft keywords for E27 rules (#11446)
• [pyflakes] Recommend adding unused import bindings to __all__ (#11314)
• [pyflakes] Update documentation and deprecate ignore_init_module_imports (#11436)
• [pyupgrade] Mark quotes as unnecessary for non-evaluated annotations (#11485)

Formatter

• Avoid multiline quotes warning with quote-style = preserve (#11490)

Server

• Support Jupyter Notebook files (#11206)
• Support noqa comment code actions (#11276)
• Fix automatic configuration reloading (#11492)
• Fix several issues with configuration in Neovim and Helix (#11497)

CLI

• Add --output-format as a CLI option for ruff config (#11438)

Bug fixes

• Avoid PLE0237 for property with setter (#11377)
• Avoid TCH005 for if stmt with elif/else block (#11376)
• Avoid flagging __future__ annotations as required for non-evaluated type annotations (#11414)
• Check for ruff executable in 'bin' directory as installed by 'pip install --target'. (#11450)
• Sort edits prior to deduplicating in quotation fix (#11452)
• Treat escaped newline as valid sequence (#11465)
• [flake8-pie] Preserve parentheses in unnecessary-dict-kwargs (#11372)
• [pylint] Ignore __slots__ with dynamic values (#11488)
• [pylint] Remove try body from branch counting (#11487)
• [refurb] Respect operator precedence in FURB110 (#11464)

Documentation

• Add --preview to the README (#11395)
• Add Python 3.13 to list of allowed Python versions (#11411)
• Simplify Neovim setup documentation (#11489)
• Update CONTRIBUTING.md to reflect the new parser (#11434)
• Update server documentation with new migration guide (#11499)
• [pycodestyle] Clarify motivation for E713 and E714 (#11483)
• [pyflakes] Update docs to describe WAI behavior (F541) (#11362)
• [pylint] Clearly indicate what is counted as a branch (#11423)

Contributors

• @AlexWaygood
• @DimitriPapadopoulos
• @augustelalande
• @charliermarsh
• @dhruvmanila
• @dougthor42
• @ekohilas
• @github-actions
• @jaap3
• @jaraco
• @konstin
• @lemeb
• @mnme
• @plredmond
• @renovate
• @rmodpur
• @snowsignal
• @thatch
• @tjkuson
• @zanieb

v0.4.4

Changes

Preview features

• [pycodestyle] Ignore end-of-line comments when determining blank line rules (#11342)
• [pylint] Detect pathlib.Path.open calls in unspecified-encoding (PLW1514) (#11288)
• [flake8-pyi] Implement PYI059 (generic-not-last-base-class) (#11233)
• [flake8-pyi] Implement PYI062 (duplicate-literal-member) (#11269)

Rule changes

• [flake8-boolean-trap] Allow passing booleans as positional-only arguments in code such as set(True) (#11287)
• [flake8-bugbear] Ignore enum classes in cached-instance-method (B019) (#11312)

Server

• Expand tildes when resolving Ruff server configuration file (#11283)
• Fix ruff server hanging after Neovim closes (#11291)
• Editor settings are used by default if no file-based configuration exists (#11266)

Bug fixes

• [pylint] Consider with statements for too-many-branches (PLR0912) (#11321)
• [flake8-blind-except, tryceratops] Respect logged and re-raised expressions in nested statements (BLE001, TRY201) (#11301)
• Recognise assignments such as __all__ = builtins.list(["foo", "bar"]) as valid __all__ definitions (#11335)

Contributors

• @Abdur-rahmaanJ
• @AlexWaygood
• @KPCOFGS
• @MichaReiser
• @augustelalande
• @blueraft
• @carljm
• @carloshbcabral
• @charliermarsh
• @dhruvmanila
• @renovate
• @snowsignal
• @tusharsadhwani

v0.4.3

Changes

Enhancements

• Add support for PEP 696 syntax (#11120)

Preview features

• [refurb] Use function range for reimplemented-operator diagnostics (#11271)
• [refurb] Ignore methods in reimplemented-operator (FURB118) (#11270)
• [refurb] Implement fstring-number-format (FURB116) (#10921)
• [ruff] Implement redirected-noqa (RUF101) (#11052)
• [pyflakes] Distinguish between first-party and third-party imports for fix suggestions (#11168)

Rule changes

• [flake8-bugbear] Ignore non-abstract class attributes when enforcing B024 (#11210)
• [flake8-logging] Include inline instantiations when detecting loggers (#11154)
• [pylint] Also emit PLR0206 for properties with variadic parameters (#11200)
• [ruff] Detect duplicate codes as part of unused-noqa (RUF100) (#10850)

Formatter

• Avoid multiline expression if format specifier is present (#11123)

LSP

• Write ruff server setup guide for Helix (#11183)
• ruff server no longer hangs after shutdown (#11222)
• ruff server reads from a configuration TOML file in the user configuration directory if no local configuration exists (#11225)
• ruff server respects per-file-ignores configuration (#11224)
• ruff server: Support a custom TOML configuration file (#11140)
• ruff server: Support setting to prioritize project configuration over editor configuration (#11086)

Bug fixes

• Avoid debug assertion around NFKC renames (#11249)
• [pyflakes] Prioritize redefined-while-unused over unused-import (#11173)
• [ruff] Respect async expressions in comprehension bodies (#11219)
• [pygrep_hooks] Fix blanket-noqa panic when last line has noqa with no newline (PGH004) (#11108)
• [perflint] Ignore list-copy recommendations for async for loops (#11250)
• [pyflakes] Improve invalid-print-syntax documentation (#11171)

Performance

• Avoid allocations for isort module names (#11251)
• Build a separate ARM wheel for macOS (#11149)

Windows

• Increase the minimum requirement to Windows 10.

Contributors

• @AlexWaygood
• @JelleZijlstra
• @KPCOFGS
• @MichaReiser
• @augustelalande
• @carljm
• @charliermarsh
• @dhruvmanila
• @diceroll123
• @ibraheemdev
• @plredmond
• @renovate
• @snowsignal
• @tusharsadhwani

v0.4.2

Changes

Rule changes

• [flake8-pyi] Allow for overloaded __exit__ and __aexit__ definitions (PYI036) (#11057)
• [pyupgrade] Catch usages of "%s" % var and provide an unsafe fix (UP031) (#11019)
• [refurb] Implement new rule that suggests min/max over sorted() (FURB192) (#10868)

Server

• Fix an issue with missing diagnostics for Neovim and Helix (#11092)
• Implement hover documentation for noqa codes (#11096)
• Introduce common Ruff configuration options with new server settings (#11062)

Bug fixes

• Use macos-12 for building release wheels to enable macOS 11 compatibility (#11146)
• [flake8-blind-expect] Allow raise from in BLE001 (#11131)
• [flake8-pyi] Allow simple assignments to None in enum class scopes (PYI026) (#11128)
• [flake8-simplify] Avoid raising SIM911 for non-zip attribute calls (#11126)
• [refurb] Avoid operator.itemgetter suggestion for single-item tuple (#11095)
• [ruff] Respect per-file-ignores for RUF100 with no other diagnostics (#11058)
• [ruff] Fix async comprehension false positive (RUF029) (#11070)

Documentation

• [flake8-bugbear] Document explicitly disabling strict zip (B905) (#11040)
• [flake8-type-checking] Mention lint.typing-modules in TCH001, TCH002, and TCH003 (#11144)
• [isort] Improve documentation around custom isort sections (#11050)
• [pylint] Fix documentation oversight for invalid-X-returns (#11094)

Performance

• Use matchit to resolve per-file settings (#11111)

Contributors

• @AlexWaygood
• @JonathanPlasse
• @KPCOFGS
• @KotlinIsland
• @MichaReiser
• @augustelalande
• @autinerd
• @bersbersbers
• @carljm
• @charliermarsh
• @dhruvmanila
• @ibraheemdev
• @jfrost-mo
• @mistercrunch
• @nolanking90
• @ottaviohartman
• @plredmond
• @renovate
• @snowsignal

v0.4.1

Changes

Preview features

• [pylint] Implement invalid-hash-returned (PLE0309) (#10961)
• [pylint] Implement invalid-index-returned (PLE0305) (#10962)

Bug fixes

• [pylint] Allow NoReturn-like functions for __str__, __len__, etc. (PLE0307) (#11017)
• Parser: Use empty range when there's "gap" in token source (#11032)
• [ruff] Ignore stub functions in unused-async (RUF029) (#11026)
• Parser: Expect indented case block instead of match stmt (#11033)

Contributors

• @AlexWaygood
• @HenryAsa
• @MithicSpirit
• @charliermarsh
• @dhruvmanila
• @tibor-reiss

v0.4.0

Changes

A new, hand-written parser

Ruff's new parser is >2x faster, which translates to a 20-40% speedup for all linting and formatting invocations. There's a lot to say about this exciting change, so check out the blog post for more details!

See #10036 for implementation details.

A new language server in Rust

With this release, we also want to highlight our new language server. ruff server is a Rust-powered language server that comes built-in with Ruff. It can be used with any editor that supports the Language Server Protocol (LSP). It uses a multi-threaded, lock-free architecture inspired by rust-analyzer and it will open the door for a lot of exciting features. It’s also faster than our previous Python-based language server -- but you probably guessed that already.

ruff server is only in alpha, but it has a lot of features that you can try out today:

• Lints Python files automatically and shows quick-fixes when available
• Formats Python files, with support for range formatting
• Comes with commands for quickly performing actions: ruff.applyAutofix, ruff.applyFormat, and ruff.applyOrganizeImports
• Supports source.fixAll and source.organizeImports source actions
• Automatically reloads your project configuration when you change it

To setup ruff server with your editor, refer to the README.md.

Preview features

• [pycodestyle] Do not trigger E3 rules on defs following a function/method with a dummy body (#10704)
• [pylint] Implement invalid-bytes-returned (E0308) (#10959)
• [pylint] Implement invalid-length-returned (E0303) (#10963)
• [pylint] Implement self-cls-assignment (W0642) (#9267)
• [pylint] Omit stubs from invalid-bool and invalid-str-return-type (#11008)
• [ruff] New rule unused-async (RUF029) to detect unneeded async keywords on functions (#9966)

Rule changes

• [flake8-bandit] Allow urllib.request.urlopen calls with static Request argument (S310) (#10964)
• [flake8-bugbear] Treat raise NotImplemented-only bodies as stub functions (B006) (#10990)
• [flake8-slots] Respect same-file Enum subclasses (SLOT000) (#11006)
• [pylint] Support inverted comparisons (PLR1730) (#10920)

Linter

• Improve handling of builtin symbols in linter rules (#10919)
• Improve display of rules in --show-settings (#11003)
• Improve inference capabilities of the BuiltinTypeChecker (#10976)
• Resolve classes and functions relative to script name (#10965)
• Improve performance of RuleTable::any_enabled (#10971)

Server

This section is devoted to updates for our new language server, written in Rust.

• Enable ruff-specific source actions (#10916)
• Refreshes diagnostics for open files when file configuration is changed (#10988)
• Important errors are now shown as popups (#10951)
• Introduce settings for directly configuring the linter and formatter (#10984)
• Resolve configuration for each document individually (#10950)
• Write a setup guide for Neovim (#10987)

Configuration

• Add RUFF_OUTPUT_FILE environment variable support (#10992)

Bug fixes

• Avoid non-augmented-assignment for reversed, non-commutative operators (PLR6104) (#10909)
• Limit commutative non-augmented-assignments to primitive data types (PLR6104) (#10912)
• Respect per-file-ignores for RUF100 on blanket # noqa (#10908)
• Consider if expression for parenthesized with items parsing (#11010)
• Consider binary expr for parenthesized with items parsing (#11012)
• Reset FOR_TARGET context for all kinds of parentheses (#11009)

Contributors

• @AlexWaygood
• @MichaReiser
• @Philipp-Thiel
• @StevenMia
• @charliermarsh
• @dhruvmanila
• @diceroll123
• @hartwork
• @hoel-bagard
• @max-muoto
• @plredmond
• @renovate
• @sisp
• @snowsignal
• @tibor-reiss
• @zanieb

v0.3.7

Changes

Preview features

• [flake8-bugbear] Implement loop-iterator-mutation (B909) (#9578)
• [pylint] Implement rule to prefer augmented assignment (PLR6104) (#9932)

Bug fixes

• Avoid TOCTOU errors in cache initialization (#10884)
• [pylint] Recode nan-comparison rule to W0177 (#10894)
• [pylint] Reverse min-max logic in if-stmt-min-max (#10890)

Contributors

• @carljm
• @charliermarsh
• @lshi18
• @mimre25

v0.3.6

Changes

Preview features

• [pylint] Implement bad-staticmethod-argument (PLW0211) (#10781)
• [pylint] Implement if-stmt-min-max (PLR1730, PLR1731) (#10002)
• [pyupgrade] Replace str,Enum multiple inheritance with StrEnum UP042 (#10713)
• [refurb] Implement if-expr-instead-of-or-operator (FURB110) (#10687)
• [refurb] Implement int-on-sliced-str (FURB166) (#10650)
• [refurb] Implement write-whole-file (FURB103) (#10802)
• [refurb] Support itemgetter in reimplemented-operator (FURB118) (#10526)
• [flake8_comprehensions] Add sum/min/max to unnecessary comprehension check (C419) (#10759)

Rule changes

• [pydocstyle] Require capitalizing docstrings where the first sentence is a single word (D403) (#10776)
• [pycodestyle] Ignore annotated lambdas in class scopes (E731) (#10720)
• [flake8-pyi] Various improvements to PYI034 (#10807)
• [flake8-slots] Flag subclasses of call-based typing.NamedTuples as well as subclasses of collections.namedtuple() (SLOT002) (#10808)
• [pyflakes] Allow forward references in class bases in stub files (F821) (#10779)
• [pygrep-hooks] Improve blanket-noqa error message (PGH004) (#10851)

CLI

• Support FORCE_COLOR env var (#10839)

Configuration

• Support negated patterns in [extend-]per-file-ignores (#10852)

Bug fixes

• [flake8-import-conventions] Accept non-aliased (but correct) import in unconventional-import-alias (ICN001) (#10729)
• [flake8-quotes] Add semantic model flag when inside f-string replacement field (#10766)
• [pep8-naming] Recursively resolve TypeDicts for N815 violations (#10719)
• [flake8-quotes] Respect Q00* ignores in flake8-quotes rules (#10728)
• [flake8-simplify] Show negated condition in needless-bool diagnostics (SIM103) (#10854)
• [ruff] Use within-scope shadowed bindings in asyncio-dangling-task (RUF006) (#10793)
• [flake8-pytest-style] Fix single-tuple conversion in pytest-parametrize-values-wrong-type (PT007) (#10862)
• [flake8-return] Ignore assignments to annotated variables in unnecessary-assign (RET504) (#10741)
• [refurb] Do not allow any keyword arguments for read-whole-file in rb mode (FURB101) (#10803)
• [pylint] Don't recommend decorating staticmethods with @singledispatch (PLE1519, PLE1520) (#10637)
• [pydocstyle] Use section name range for all section-related docstring diagnostics (#10740)
• Respect # noqa directives on __all__ openers (#10798)

Contributors

• @AlexWaygood
• @Boshen
• @MichaReiser
• @NotWearingPants
• @Warchant
• @alex-700
• @augustelalande
• @buhtz
• @carljm
• @charliermarsh
• @dhruvmanila
• @diceroll123
• @renovate
• @snowsignal
• @tibor-reiss

v0.3.5

Changes

Preview features

• [pylint] Implement modified-iterating-set (E4703) (#10473)
• [refurb] Implement for-loop-set-mutations (FURB142) (#10583)
• [refurb] Implement unnecessary-from-float (FURB164) (#10647)
• [refurb] Implement verbose-decimal-constructor (FURB157) (#10533)

Rule changes

• [flake8-comprehensions] Handled special case for C401 which also matches C416 (#10596)
• [flake8-pyi] Mark unaliased-collections-abc-set-import fix as "safe" for more cases in stub files (PYI025) (#10547)
• [numpy] Add row_stack to NumPy 2.0 migration rule (#10646)
• [pycodestyle] Allow cell magics before an import (E402) (#10545)
• [pycodestyle] Avoid blank line rules for the first logical line in cell (#10291)

Configuration

• Respected nested namespace packages (#10541)
• [flake8-boolean-trap] Add setting for user defined allowed boolean trap (#10531)

Bug fixes

• Correctly handle references in __all__ definitions when renaming symbols in autofixes (#10527)
• Track ranges of names inside __all__ definitions (#10525)
• [flake8-bugbear] Avoid false positive for usage after continue (B031) (#10539)
• [flake8-copyright] Accept commas in default copyright pattern (#9498)
• [flake8-datetimez] Allow f-strings with %z for DTZ007 (#10651)
• [flake8-pytest-style] Fix PT014 autofix for last item in list (#10532)
• [flake8-quotes] Ignore Q000, Q001 when string is inside forward ref (#10585)
• [isort] Always place non-relative imports after relative imports (#10669)
• [isort] Respect Unicode characters in import sorting (#10529)
• [pyflakes] Fix F821 false negatives when from __future__ import annotations is active (attempt 2) (#10524)
• [pyflakes] Make unnecessary-lambda an always-unsafe fix (#10668)
• [pylint] Fixed false-positive on the rule PLW1641 (eq-without-hash) (#10566)
• [ruff] Fix panic in unused # noqa removal with multi-byte space (RUF100) (#10682)

Documentation

• Add PR title format to CONTRIBUTING.md (#10665)
• Fix list markup to include blank lines required (#10591)
• Put flake8-logging next to the other flake8 plugins in registry (#10587)
• [flake8-bandit] Update warning message for rule S305 to address insecure block cipher mode use (#10602)
• [flake8-bugbear] Document use of anonymous assignment in useless-expression (#10551)
• [flake8-datetimez] Clarify error messages and docs for DTZ rules (#10621)
• [pycodestyle] Use same before vs. after numbers for space-around-operator (#10640)
• [ruff] Change quadratic-list-summation docs to use iadd consistently (#10666)

Contributors

• @AlexWaygood
• @FFY00
• @MichaReiser
• @alex-700
• @augustelalande
• @aureliojargas
• @boolean-light
• @charliermarsh
• @dhruvmanila
• @diceroll123
• @dizzy57
• @dopplershift
• @hoel-bagard
• @mnixry
• @mtsokol
• @pjonsson
• @renovate
• @snowsignal
• @yt2b
• @zanieb

v0.3.4

Changes

Preview features

• [flake8-simplify] Detect implicit else cases in needless-bool (SIM103) (#10414)
• [pylint] Implement nan-comparison (PLW0117) (#10401)
• [pylint] Implement nonlocal-and-global (E115) (#10407)
• [pylint] Implement singledispatchmethod-function (PLE5120) (#10428)
• [refurb] Implement list-reverse-copy (FURB187) (#10212)

Rule changes

• [flake8-pytest-style] Add automatic fix for pytest-parametrize-values-wrong-type (PT007) (#10461)
• [pycodestyle] Allow SPDX license headers to exceed the line length (E501) (#10481)

Formatter

• Fix unstable formatting for trailing subscript end-of-line comment (#10492)

Bug fixes

• Avoid code comment detection in PEP 723 script tags (#10464)
• Avoid incorrect tuple transformation in single-element case (C409) (#10491)
• Bug fix: Prevent fully defined links name from being reformatted (#10442)
• Consider raw source code for W605 (#10480)
• Docs: Link inline settings when not part of options section (#10499)
• Don't treat annotations as redefinitions in .pyi files (#10512)
• Fix E231 bug: Inconsistent catch compared to pycodestyle, such as when dict nested in list (#10469)
• Fix pylint upstream categories not showing in docs (#10441)
• Add missing Options references to blank line docs (#10498)
• 'Revert "F821: Fix false negatives in .py files when from __future__ import annotations is active (#10362)"' (#10513)
• Apply NFKC normalization to unicode identifiers in the lexer (#10412)
• Avoid failures due to non-deterministic binding ordering (#10478)
• [flake8-bugbear] Allow tuples of exceptions (B030) (#10437)
• [flake8-quotes] Avoid syntax errors due to invalid quotes (Q000, Q002) (#10199)

Contributors

• @AlexWaygood
• @MichaReiser
• @VascoSch92
• @WindowGenerator
• @alex-700
• @augustelalande
• @autinerd
• @boolean-light
• @charliermarsh
• @dhruvmanila
• @dopplershift
• @hoel-bagard
• @ottaviohartman
• @robincaloudis
• @snowsignal
• @trag1c
• @veryyet