Skip to content

perf(parser): make parser and lexer checkpoints Copy - #29662

Merged
Boshen merged 1 commit into
mainfrom
perf/parser-copy-checkpoint
Nov 16, 2026
Merged

Boshen merged 1 commit into
mainfrom
perf/parser-copy-checkpoint

Conversation

@Boshen

@Boshen Boshen commented Nov 16, 2026

Copy link
Copy Markdown
Member

ParserCheckpoint and LexerCheckpoint are now Copy, so taking one and dropping it costs nothing beyond copying a few words.

  • LexerCheckpoint records the number of lexer errors instead of ErrorSnapshot (Empty / Count / Full(Vec)).
  • ParserCheckpoint records whether a fatal error is set instead of taking it. A fatal error set after the checkpoint is discarded on rewind.
  • checkpoint_with_error_recovery is removed. The error vector was only cloned because set_unexpected pops a lexer error into the fatal error. rewind now gives that error back to the lexer instead.
  • set_unexpected no longer pops a lexer error when a fatal error is already set, because that error was dropped.

The count-based rewind used by checkpoint() also dropped a lexer error that set_unexpected had popped. Those errors are now reported. This updates 4 test262 hashbang entries: Expected in becomes the real invalid-escape error.


AI assistance (Claude) was used to write this change; I reviewed it.

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 16, 2026 •

Copy link
Copy Markdown
Codex Review Summary: βœ… Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
πŸ“ Code Review βœ… Completed 2026-11-16T08:41:55.864801Z be279b5 PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with πŸ‘€ while any review is running, comments if it has suggestions, and reacts with πŸ‘ once all reviews finish with no findings.

@github-actions github-actions Bot added the A-parser Area - JS/TS parsing and syntax errors, oxc_parser label Nov 16, 2026
@codspeed

codspeed Bot commented Nov 16, 2026

Copy link
Copy Markdown
Contributor

Merging this PR will not alter performance

βœ… 91 untouched benchmarks
⏩ 19 skipped benchmarks1


Comparing perf/parser-copy-checkpoint (be279b5) with main (fe2e300)

Open in CodSpeed

Footnotes

  1. 19 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

@Boshen Boshen added the 0-merge Merge with Graphite Merge Queue label Nov 16, 2026

Boshen commented Nov 16, 2026 •

Copy link
Copy Markdown
Member Author

Merge activity

  • Nov 16, 8:53 AM UTC: The merge label '0-merge' was detected. This PR will be added to the Graphite merge queue once it meets the requirements.
  • Nov 16, 8:53 AM UTC: Boshen added this pull request to the Graphite merge queue.

@Boshen
Boshen merged commit dd9af34 into main Nov 16, 2026
43 checks passed
@Boshen
Boshen deleted the perf/parser-copy-checkpoint branch November 16, 2026 08:54
graphite-app Bot pushed a commit that referenced this pull request Nov 16, 2026
### πŸ’₯ BREAKING CHANGES

- d742f8b ast: [**BREAKING**] Add `LabelId` to `LabelIdentifier` (#29611) (camc314)

### πŸš€ Features

- 63b8158 minifier: Inline single use function declaration to assignment (#28917) (Aadharsh  Pannirselvam)
- eb1833e mangler: Mangle labels (#29614) (camc314)
- 702e734 codegen: Print label names from scoping (#29613) (camc314)
- 750033a semantic: Resolve labels to label IDs (#29612) (camc314)
- ccca379 minifier: Remove unused labels (#29607) (camc314)
- e375dff lexer: Complete NEON support for aarch64 (#29390) (Shanon Jackson)
- bfe9a81 data_structures: Add buffered vector batch reservations (#29365) (Boshen)
- fa492d6 minifier: Merge adjacent switch cases with identical bodies (#29316) (camc314)
- 8cc6101 minifier: Inline member reads into callees as indirect calls (#29305) (camc314)
- e3d7d91 minifier: Inline member reads into `new` expressions (#29304) (camc314)
- 57211d0 minifier: Inline member reads into call arguments (#29303) (camc314)
- 523a063 lexer: Publish lexer and SIMD crates (#29264) (Boshen)
- 2a5d710 data_structures: Add `unreachable_unchecked!` macro (#29131) (overlookmotel)
- 061becc react-compiler: Support logical assignment operators (#29137) (Yagiz Nizipli)

### πŸ› Bug Fixes

- d520358 transformer: Retain decorated fields with define semantics (#29444) (Dunqing)
- 83fb644 lexer: Combine surrogate pairs written with any escape form (#29658) (Shanon Jackson)
- f8e329c transformer: Preserve optional calls on private field chains (#29443) (Dunqing)
- 4b86eba react-compiler: Do not report hoisted function declaration initializer as access before declaration (#29641) (Boshen)
- 7e41638 parser: Keep whole parenthesized `TSInstantiationExpression` with `preserve_parens: false` (#29359) (leaysgur)
- 5aa6c17 napi/parser: Release native parse result eagerly (#29569) (Boshen)
- 1dc0ee9 semantic: Preserve catch and var binding identities (#28739) (Dunqing)
- 11dbaf4 ecmascript: Prefer hex integers when lengths tie (#29436) (camc314)
- e64155e transformer/react-refresh: Recognize TypeScript wrappers in HOC arguments (#29290) (camc314)
- 9085be1 react-compiler: Only report zero-argument new Date in render body (#29419) (Boshen)
- 9f465fd react-compiler: Preserve ref mutation hints (#28553) (Nash Aschenbath)
- 075b74f lexer: Do not drop UTF-8 validation errors in TSX type arguments (#29409) (overlookmotel)
- 59cbbbf lexer: Fix source length assertion (#29408) (overlookmotel)
- f37f271 react-compiler: Resolve computed dependencies in merged memo scopes (#29376) (camc314)
- ab73590 parser: Preserve re-exported default import names in raw transfer (#29234) (camc314)
- 6e6bd61 react-compiler: Avoid invariants for catch phis and ref-free callbacks (#29018) (camc314)
- 19b1691 isolated-declarations: Retain types used by exported ambient variables (#28996) (camc314)
- b372b04 transformer/optional-chaining: Preserve parenthesized chain boundaries (#29218) (camc314)
- efcdf06 transformer: Isolate logical assignment temporaries in parameters (#29182) (Dunqing)
- 177fdd7 parser: Classify re-exported namespaces as indirect exports (#29196) (Boshen)
- 053cc40 parser: Preserve type-only flags on re-exported imports (#29195) (Boshen)
- 34172e0 data_structures: Fix `assert_unchecked!` macro (#29130) (overlookmotel)
- b515929 transformer: Lower wrapped assignments and preserve computed receivers (#29113) (camc314)
- 59aed23 parser: Highlight invalid numeric separators (#29117) (camc314)
- 8f5e82c parser: Reject exponentiation operands hidden by assertions (#29115) (camc314)
- 42e790c parser: Require `declare` before `abstract` modifiers (#29111) (camc314)

### ⚑ Performance

- dd9af34 parser: Make parser and lexer checkpoints `Copy` (#29662) (Boshen)
- 4ebe523 parser: Keep the tokens read while classifying `()` and `(x)` arrows (#29560) (Boshen)
- 7a6b0a4 parser: Lex the token after `let` and `?.` once (#29559) (Boshen)
- bd9fa9b parser: Lex the token after a modifier keyword once (#29558) (Boshen)
- 2c6ee72 parser: Skip arrow speculation for array and object literals (#29557) (Boshen)
- 85c0269 parser: Rule out type arguments from tokens before speculating (#29556) (Boshen)
- b6edb78 parser: Avoid token loop restart after newlines (#29423) (Boshen)
- 762bbcf lexer: Cut NEON lexing cost in carve, coalesce and compress (#29414) (Shanon Jackson)
- f09af45 lexer: Keep source position in registers across byte handlers (#29401) (Boshen)
- ae1a375 lexer: Complete single-byte tokens without a byte handler (#29400) (Boshen)
- 11daf98 lexer: Build tokens once from pending state (#29399) (Boshen)
- c49e1bd str: Hash identifiers without branching on length (#29397) (Boshen)
- 8a6e0c1 lexer: Borrow contiguous escaped string values (#29372) (Boshen)
- 27baf0c lexer: Buffer more comments before arena allocation (#29371) (Boshen)
- baa84fc lexer: Size escaped strings before allocating (#29369) (Boshen)
- 4313dee parser: Buffer module record entries (#29368) (Boshen)
- 2df1f19 parser: Reduce regex, template, and comment allocations (#29361) (Boshen)
- f26d9d9 parser: Reduce arena allocations and reallocations (#29358) (Boshen)
- 846adbb mangler: Skip a liveness walk that never marks a scope (#29391) (Dunqing)
- 0d2e41a parser: Reduce identifier dispatch overhead (#29347) (Boshen)
- 1aaf3b0 parser: Reuse pure re-export storage (#29216) (Boshen)
- 9add80d parser: Buffer decorator and heritage lists (#29215) (Boshen)
- cb5cf55 parser: Collect prefixed import lists once (#29214) (Boshen)
- ec6b166 parser: Omit discarded recovered parentheses (#29213) (Boshen)
- 01e2b66 parser: Avoid discarded TypeScript instantiations (#29210) (Boshen)
- 727610a parser: Collect short named import lists (#29209) (Boshen)
- d1eaed0 parser: Collect short template literal lists (#29208) (Boshen)
- 0301b92 parser: Reserve module record batches (#29207) (Boshen)
- 9734804 parser: Collect short switch consequent lists (#29206) (Boshen)
- 6674b7a parser: Collect short JSX child lists (#29205) (Boshen)
- cc03246 parser: Collect short binding pattern lists (#29204) (Boshen)
- e36dd2d parser: Avoid discarded cover parameter allocations (#29203) (Boshen)
- 2d5864c parser: Collect short sequence expression lists (#29202) (Boshen)
- 6495e66 parser: Avoid discarded block allocations (#29199) (Boshen)
- 2601c45 parser: Collect short formal parameter lists (#29198) (Boshen)
- 10fddc3 parser: Collect short TypeScript type lists (#29197) (Boshen)
- 4140fb0 parser: Keep class heritage on the stack (#29190) (Boshen)
- e028d74 parser: Collect short statement lists before allocating (#29189) (Boshen)
- 69c7892 parser: Collect short declaration lists before allocating (#29188) (Boshen)
- aed0190 parser: Skip unused identifier hashes (#29168) (Boshen)
- 2c0a6c1 parser: Avoid allocating discarded directive statements (#29186) (Boshen)
- 48532bb parser: Buffer remaining parser lists (#29185) (Boshen)
- e59d413 parser: Reuse export entry storage (#29172) (Boshen)
- f815f54 parser: Reduce arena list reallocations (#29169) (Boshen)

### πŸ“š Documentation

- 9e5a2bb agents: Require runnable playground reproductions in PRs (#29450) (Dunqing)
- c4d1024 minifier: Require examples in fix and optimization PRs (#29439) (Dunqing)
- 43e0829 parser: Fix grammar context formatting (#29373) (camc314)
- 3e947d7 napi/parser: Correct TS-ESTree differences (#29268) (fisker Cheung)
- 521d194 transformer: Require reproductions in fix PR descriptions (#29224) (Dunqing)
- 24c4b78 simd: Improve comment (#29244) (overlookmotel)
- 306b96a lexer: Correct safety comment on `TokensView::new` (#29231) (overlookmotel)
- 44c65f0 lexer: Add comments (#29226) (overlookmotel)
- 796255d simd: Use consistent formatting for bullet points in doc comments (#29133) (overlookmotel)
- 794cf47 isolated-declarations: Add agent guide (#29125) (camc314)
- 768cc29 parser: Clarify syntax support and AST preservation (#29121) (camc314)
- 67ecd8c estree: Correct serialization capabilities (#29120) (camc314)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

0-merge Merge with Graphite Merge Queue A-parser Area - JS/TS parsing and syntax errors, oxc_parser

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant