fallintoplace opened a new pull request, #1949: URL: https://github.com/apache/iceberg-go/pull/1949
## What changed - **Precompute Parquet column mappings once per writer lifecycle.** - Reuse variant field IDs too. - Pass both maps through the default and rolling writer paths. - Keep the existing fallback for direct file-writer callers. - Add correctness coverage and a writer setup benchmark. ## Why Each physical Parquet file rebuilt the same schema mappings even though the schema is invariant for the writer. The maps are read-only, so they can be shared safely across files. ## Checks - `go test ./... -count=1` - `go test -race ./table -count=1` - `go test -race ./table/internal -count=1` - `go vet ./table ./table/internal` The focused benchmark shows about 12-14% lower writer setup time and 33-790 fewer allocations per file for 8-256 fields. -- 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]
