You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@paimon.apache.org by "gripleaf (via GitHub)" <gi...@apache.org> on 2026/09/08 13:38:55 UTC

[PR] perf(parquet): avoid duplicate pre-buffering for mixed row groups [paimon-cpp]

gripleaf opened a new pull request, #304:
URL: https://github.com/apache/paimon-cpp/pull/304

   ### Purpose
   
     When a scan includes both fully matched and page-filtered row groups, Arrow’s automatic pre-buffering and Paimon’s range pre-buffering can read the same data twice.
   
     Let FileReaderWrapper manage a shared pre-buffer cache for both kinds of row groups. Refresh cache coverage when seeking, preserve Arrow’s standard handling for full-only scans, and retain existing fallback behavior.
   
     ### Tests
   
     Added regression tests in FileReaderWrapperTest:
   
     - ManagedPreBufferUsesSingleCacheForMixedRowGroups: checks equivalent results with fewer bytes read for mixed scans and unchanged I/O for full-only and partial-only scans.
     - ManagedPreBufferBackwardSeekRefreshesCacheCoverage: checks cache coverage after backward seeks.
     - ManagedPreBufferPropagatesReadErrors: checks propagation of injected read failures.
     - DisabledPreBufferPreservesPageFilteredCache: checks page-filtered caching when automatic pre-buffering is disabled.
   
     Tests were not run while preparing this description. No integration tests were added.
   
     ### API and Format
   
     No changes to public APIs under include/, storage formats, or protocols.
   
     ### Documentation
   
     No new user-facing features or configuration options. No documentation changes required.
   
     ### Generative AI tooling
   
     Generated-by: OpenAI Codex (GPT-6)


-- 
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: issues-unsubscribe@paimon.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

Re: [PR] perf(parquet): avoid duplicate pre-buffering for mixed row groups [paimon-cpp]

Posted by "gripleaf (via GitHub)" <gi...@apache.org>.
gripleaf commented on PR #304:
URL: https://github.com/apache/paimon-cpp/pull/304#issuecomment-5707212089

   Effective remote retest: QPS 77.4735, E2E mean 50.855 ms.
   Compared to PR #329: QPS +1.80%, mean improvement 2.00%.
   
   The optimization review shows limited improvement.


-- 
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: issues-unsubscribe@paimon.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

Re: [PR] perf(parquet): avoid duplicate pre-buffering for mixed row groups [paimon-cpp]

Posted by "gripleaf (via GitHub)" <gi...@apache.org>.
gripleaf closed pull request #304: perf(parquet): avoid duplicate pre-buffering for mixed row groups
URL: https://github.com/apache/paimon-cpp/pull/304


-- 
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: issues-unsubscribe@paimon.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org