XiaoHongbo-Hope opened a new pull request, #1010: URL: https://github.com/apache/paimon-rust/pull/1010
## Summary On a BLOB index cache miss, reading the footer and index currently requires two range reads. Prefetch the last 4 KiB (or the whole file if smaller) and decode the index from that buffer when it fits. Larger indexes fetch only the missing prefix. This trades a small amount of extra tail data for fewer requests, without changing the file format, cache policy, or read concurrency. Existing file sizes are reused; no HEAD request is added. ## Validation - 174 BLOB-related tests passed, including ARRAY/MAP, descriptor offsets, row selection, and cache hits. - Added boundary, malformed index, short-read, and I/O failure tests. - A deterministic local 16-file cold-read comparison reduced FileRead calls from 32 to 16, with identical decoded entries; bytes increased from 24,616 to 65,536. These are synthetic request counts, not OSS or training throughput measurements. - `cargo clippy --locked -p paimon --lib --tests -- -D warnings`, `cargo fmt --all -- --check`, and `git diff --check` passed. -- 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]
