XiaoHongbo-Hope opened a new pull request, #9133: URL: https://github.com/apache/paimon/pull/9133
### Purpose Coalesced BLOB range reads currently open and close an input stream for every merged span. A batch containing several spans from the same physical file therefore repeats stream-open overhead even though positional reads can share one stream. ### Changes - Reuse one lazily opened input stream for all coalesced spans of the same path. - Run positional reads concurrently and serialize seek/read for streams without positional reads. - Fall back to an independently opened range when a shared-stream read fails. - Close a path stream as soon as its last task completes, keeping open resources bounded by active work. - Forward the range-stream hook through FileIO wrappers. - Use the native Jindo stream so shared reads call pread directly. The direct FormatBlobReader path is unchanged. ### Tests - Verify 64 separate spans from one URI use one open, 64 positional reads, and one close. - Verify non-positional streams serialize reads. - Verify shared-stream failures fall back to independent reads. - Verify FileIO wrappers and the Jindo adapter preserve the range-stream path. - Run the relevant Blob, multimodal, FileIO, REST, and Jindo tests: 244 passed, 21 skipped, 45 subtests 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]
