escape: rewrite InternalEscapeBytes with lead-byte scanning #39
Open
dhartunian wants to merge 2 commits intomasterfrom
Open
escape: rewrite InternalEscapeBytes with lead-byte scanning #39dhartunian wants to merge 2 commits intomasterfrom
dhartunian wants to merge 2 commits intomasterfrom
Conversation
Add test cases for edge cases including truncated lead bytes, non-marker UTF-8 sequences sharing the same lead byte (e.g. em dash, euro sign, ellipsis), trailing invalid UTF-8, multiple consecutive markers, and interleaved markers with newlines. Also add benchmarks exercising redactable string printing with interpolated SafeFormatter structs. Co-Authored-By: roachdev-claude <roachdev-claude-bot@cockroachlabs.com>
InternalEscapeBytes is called on every buffer mode transition during redactable string formatting, scanning newly-written bytes for marker characters (‹ › †) that must be escaped. Format operations like Sprintf with SafeFormatter structs trigger many mode transitions, making this function a significant cost in redactable output paths. The previous implementation compared every byte against each 3-byte marker using bytes.Equal. Since all markers share the same first two UTF-8 bytes (0xE2 0x80), replace the per-byte scan with bytes.IndexByte to skip ASCII runs in bulk (leveraging SIMD on arm64/amd64), then check only the third byte to distinguish markers. For the breakNewLines path, use dual bytes.IndexByte calls to find either the lead byte or newline. benchdiff --old 8f1ddc0 --new 82c8dd4 --count 10 --run BenchmarkSprintfWithSafeFormatter . name old time/op new time/op delta SprintfWithSafeFormatter/multiple_structs-10 1.85µs ± 2% 1.25µs ± 2% -32.34% (p=0.000 n=8+9) SprintfWithSafeFormatter/nested_structs-10 1.01µs ± 1% 0.65µs ± 4% -35.40% (p=0.000 n=8+9) SprintfWithSafeFormatter/single_struct-10 517ns ± 1% 328ns ± 1% -36.50% (p=0.000 n=9+8) SprintfWithSafeFormatter/sprint_mixed-10 1.80µs ± 2% 1.29µs ± 3% -28.52% (p=0.000 n=9+9) Co-Authored-By: roachdev-claude <roachdev-claude-bot@cockroachlabs.com>
Contributor
Author
|
Not sure yet if I'll merge this but putting it up since it's an interesting result |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
InternalEscapeBytes is called on every buffer mode transition during
redactable string formatting, scanning newly-written bytes for marker
characters (‹ › †) that must be escaped. Format operations like
Sprintf with SafeFormatter structs trigger many mode transitions,
making this function a significant cost in redactable output paths.
The previous implementation compared every byte against each 3-byte
marker using bytes.Equal. Since all markers share the same first two
UTF-8 bytes (0xE2 0x80), replace the per-byte scan with
bytes.IndexByte to skip ASCII runs in bulk (leveraging SIMD on
arm64/amd64), then check only the third byte to distinguish markers.
For the breakNewLines path, use dual bytes.IndexByte calls to find
either the lead byte or newline.
Co-Authored-By: roachdev-claude roachdev-claude-bot@cockroachlabs.com
This change is