baiyangtx commented on PR #7866:
URL: https://github.com/apache/paimon/pull/7866#issuecomment-4515909075

   > Hi @baiyangtx @Aitozi , please take a look to #7842 , I didn't look 
closely, maybe the abilities of these two PR implementations are the same.
   
   Thanks @umi for working on this too. My PR #7866 takes a different approach 
— sorting inline during the existing full compaction path, rather than a 
separate post-commit sort pipeline. Key differences:
   
   - **No new compact type** — sorting happens inside 
`ManifestFileMerger.tryFullCompaction`, reusing the existing full compaction 
trigger and lifecycle. No additional manifest scans or rewrite passes.
   - **Memory-safe with external sort** — uses `BinaryExternalSortBuffer` 
(already battle-tested across Paimon). When entries exceed 
`manifest.merge.sort.buffer`, it automatically spills to disk. No risk of OOM 
regardless of manifest count.
   - **Finer-grained config** — separate toggles for merge compaction, 
commit-time merge, and delta writing, instead of a single switch.
   
   Happy to align with the community's direction, but I believe the inline 
approach is simpler and less invasive.


-- 
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]

Reply via email to