Repository navigation
Speed up identifier and keyword lexing - #29463
Conversation
Memory usage reportMemory usage unchanged β |
|
|
Merging this PR will improve performance by 4.86%1 warning
Performance Changes
Tip Curious why performance improved? Comment Comparing Footnotes
|
a1177bd to
f91d049
Compare
ea02032 to
d47bd15
Compare
d47bd15 to
03bf69f
Compare
|
why? |
|
I got the vibe from your comments that you didn't want to proceed with it haha! Do you think it's salvageable? I'm happy to clean it up, I just thought you were left feeling like the tradeoffs weren't right. |
|
I think it's a worthwhile change. I just think it requires some more explanation of what happens here and how to update that weird constant when adding/removing kewyords π |
|
Haha ok. I will clean it up. I'll ask Codex to make it even weirder and more confusing. |
| # dependencies = [] | ||
| # /// | ||
|
|
||
| """Generate the lexer's keyword table and collision-free hash multiplier. |
There was a problem hiding this comment.
How is this different from using phf?
There was a problem hiding this comment.
So, I tried phf::Map, but it seems like it's meaningfully slower because phf still runs SipHash (the main benefit seems to be that it guarantees you don't have collisions?) but these can just fit in a u64.
MichaReiser
left a comment
There was a problem hiding this comment.
Thank you. This makes sense to me. The only part that isn't clear to me (and might be worth documenting), is why we can't use pfh. I'd prefer that over a custom Python script that generates the lookup table.
6e5b67b to
6818c90
Compare
Memory usage reportMemory usage unchanged β Results for 6818c90 Β· Workflow run |
|
## Summary We give ASCII identifiers a dedicated path, avoiding Unicode bookkeeping and checking string prefixes only when a one- or two-letter name reaches a quote. Both paths share the existing batched ASCII scanner, preserving fast scanning for Unicode names with long ASCII suffixes. Stacked on #29463 to measure the incremental improvement. [CodSpeed reports](https://github.com/astral-sh/ruff/pull/29503/checks?check_run_id=118300735426) 6.6β9.8% greater efficiency across all five lexer benchmarks, with no reported regressions. We preserve the existing scanner and add no inlining attributes or unsafe code.
This PR contains the following updates: | Package | Type | Update | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | |---|---|---|---|---|---| | [DavidAnson/markdownlint-cli2](https://github.com/DavidAnson/markdownlint-cli2) | repository | minor | `v0.23.3` β `v0.24.0` |  |  | | [ruff](https://docs.astral.sh/ruff) ([source](https://github.com/astral-sh/ruff), [changelog](https://github.com/astral-sh/ruff/blob/main/CHANGELOG.md)) | dependency-groups | patch | `==0.17.1` β `==0.17.3` |  |  | Note: The `pre-commit` manager in Renovate is not supported by the `pre-commit` maintainers or community. Please do not report any problems there, instead [create a Discussion in the Renovate repository](https://github.com/renovatebot/renovate/discussions/new) if you have any questions. --- ### Release Notes <details> <summary>DavidAnson/markdownlint-cli2</summary> ### [`v0.24.0`](DavidAnson/markdownlint-cli2@v0.23.3...v0.24.0) [Compare Source](DavidAnson/markdownlint-cli2@v0.23.3...v0.24.0) </details> <details> <summary>astral-sh/ruff (ruff)</summary> ### [`v0.17.3`](https://github.com/astral-sh/ruff/blob/HEAD/CHANGELOG.md#0173) [Compare Source](astral-sh/ruff@0.17.2...0.17.3) Released on 2026-10-26. ##### Preview features - \[`suspicious`] Add `shared-clock-mock` ([#​29074](astral-sh/ruff#29074)) ##### Bug fixes - \[`pyupgrade`] Avoid incorrect conversions to f-strings (`UP032`) ([#​25485](astral-sh/ruff#25485)) ##### Performance - Buffer stdout in `ruff format` ([#​29631](astral-sh/ruff#29631)) - Specialize ASCII identifier lexing ([#​29503](astral-sh/ruff#29503)) - Speed up identifier and keyword lexing ([#​29463](astral-sh/ruff#29463)) ##### Documentation - \[`flake8-bandit`] Describe what the rule actually checks for `httpx` (`S113`) ([#​27936](astral-sh/ruff#27936)) ##### Contributors - [@​itsvedantkumar](https://github.com/itsvedantkumar) - [@​charliermarsh](https://github.com/charliermarsh) - [@​anishgirianish](https://github.com/anishgirianish) ### [`v0.17.2`](https://github.com/astral-sh/ruff/blob/HEAD/CHANGELOG.md#0172) [Compare Source](astral-sh/ruff@0.17.1...0.17.2) Released on 2026-10-22. ##### Preview features - \[`complexity`] Add `unnecessary-regular-expression-compile` ([#​25533](astral-sh/ruff#25533)) ##### Bug fixes - Avoid syntax errors for type variables with starred constraints (`UP040`, `UP046`, `UP047`, `RUF053`) ([#​28362](astral-sh/ruff#28362)) - \[`pydocstyle`] Treat backslash-escaped `*args` and `**kwargs` as documented (`D417`) ([#​29536](astral-sh/ruff#29536)) - \[`pylint`] Avoid false positive on starred elements (`PLE0643`) ([#​29483](astral-sh/ruff#29483)) - \[`pylint`] Mark fixes unsafe when deleting comments (`PLR5501`, `PLW0120`) ([#​29414](astral-sh/ruff#29414)) - \[`pyupgrade`] Mark fix unsafe when deleting comments (`UP035`) ([#​29413](astral-sh/ruff#29413)) ##### Performance - Avoid intermediate names in lint name resolution ([#​29506](astral-sh/ruff#29506)) - Prefilter bidirectional Unicode checks by UTF-8 leading bytes ([#​29561](astral-sh/ruff#29561)) - Share postfix parsing and reduce expression copies ([#​29480](astral-sh/ruff#29480)) ##### Configuration - \[`pylint`] Add a configurable allowlist for `magic-value-comparison` (`PLR2004`) ([#​18961](astral-sh/ruff#18961)) ##### Documentation - Clarify documentation and correct stale guidance ([#​29374](astral-sh/ruff#29374)) - \[`pyupgrade`] Remove mentions of variance related to fix safety (`UP040`, `UP047`) ([#​29592](astral-sh/ruff#29592)) ##### Contributors - [@​fixfon](https://github.com/fixfon) - [@​MichaReiser](https://github.com/MichaReiser) - [@​zaniebot](https://github.com/zaniebot) - [@​FrancescoPezzella](https://github.com/FrancescoPezzella) - [@​anishgirianish](https://github.com/anishgirianish) - [@​rajveer3x](https://github.com/rajveer3x) - [@​charliermarsh](https://github.com/charliermarsh) - [@​LckyLike](https://github.com/LckyLke) - [@​DaniBodor](https://github.com/DaniBodor) </details> --- ### Configuration π **Schedule**: (in timezone America/Denver) - Branch creation - Between 02:00 AM and 06:59 AM, only on Tuesday (`* 2-6 * * 2`) - Automerge - At any time (no schedule defined) π¦ **Automerge**: Disabled by config. Please merge this manually once you are satisfied. β» **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. π» **Immortal**: This PR will be recreated if closed unmerged. Get [config help](https://github.com/renovatebot/renovate/discussions) if that's undesired. --- - [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check this box --- This PR has been generated by [Mend Renovate CLI](https://github.com/renovatebot/renovate). <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0NC4xNDguMyIsInVwZGF0ZWRJblZlciI6IjQ0LjE0OC4zIiwidGFyZ2V0QnJhbmNoIjoibWFpbiIsImxhYmVscyI6WyJyZW5vdmF0ZSJdfQ==--> Reviewed-on: https://git.lichtenwalter.site/rlichten/dir2text/pulls/100
Summary
We scan ASCII identifier continuations in sixteen-byte batches using a byte lookup table, avoiding per-character decoding. Keyword lookup filters by length and initial byte, then uses a compact hash table with a full-word comparison to reject collisions. Non-ASCII suffixes retain the existing Unicode path.