fallintoplace opened a new pull request, #1934: URL: https://github.com/apache/iceberg-go/pull/1934
## Summary - **Avoid a full-string rune scan and `[]rune` allocation** when truncating text bounds. - Walk only to the requested UTF-8 rune boundary. - Reuse the same path for variant string lower and upper bounds. - Keep upper-bound carry behavior for the surrogate gap and maximal rune cases. - Clone only the small truncated lower bound so it does not retain a large source string. ⚡ ## Benchmark Local Apple M1 Pro. `width=16`, `-benchmem`, `-benchtime=100ms`. | Path | Before | After | | --- | ---: | ---: | | `TruncateUpperBoundText`, ASCII 64 KB | ~111,008 ns/op, 262,169 B/op | ~38 ns/op, 16 B/op | | `TruncateUpperBoundText`, UTF-8 64 KB | ~375,620 ns/op, 139,312 B/op | ~73 ns/op, 32 B/op | | Variant string bounds, ASCII 64 KB | ~285,109 ns/op, 524,432 B/op | ~169 ns/op, 128 B/op | | Variant string bounds, UTF-8 64 KB | ~624,216 ns/op, 278,721 B/op | ~236 ns/op, 160 B/op | The benchmark also covers 32 B and 1 KB values, Unicode boundaries, surrogate-gap carry, maximal-rune carry, and the all-maximal case. ## Tests - `go test ./table/... -count=1` - `go test ./... -run "^$" -count=1` - `go test ./table/internal -run "^$" -bench "^(BenchmarkTruncateUpperBoundText|BenchmarkTruncateVariantBoundString)$" -benchmem -benchtime=100ms -count=1` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
