JeonDaehong opened a new pull request, #1377:
URL: https://github.com/apache/paimon-rust/pull/1377

   ### Purpose
   
   Linked issue: none, follows #1370
   
   #1370 lets aggregation read a single-run section without the sort merge. 
Partial-update returns a lone version as is too (`singleton_reducer_result`), 
so it can take the same path, again except under `changelog-producer = lookup`.
   
   ### Brief change log
   
   - The single-run check also accepts partial-update unless 
`changelog-producer = lookup`.
   
   PK table, 2M keys in 4 files with disjoint keys, full read (release, 
in-memory FileIO): 195–196 → 89–90 ms, the same as deduplicate. Measured in one 
binary with the merge path switched back by an env var, 3 alternating rounds. 
With overlapping rounds and for deduplicate, times are unchanged.
   
   ### Tests
   
   - 
`partial_update_reads_disjoint_files_without_merge_and_overlaps_still_merge`: 
disjoint files read as written; an overlapping file with a null `v` keeps the 
earlier value.
   - 
`partial_update_sequence_group_reads_single_runs_without_its_sequence_field`: 
with `fields.g.sequence-group = v`, a read of `id, v` reads `g` internally; 
single runs read the same, and overlapping files still follow `g`.
   - `single_run_read_merges_repeated_keys_under_partial_update`: a pre-#839 
file that repeats a key still merges.
   - `single_run_read_drops_retracts_hidden_by_legacy_metadata` gets a 
partial-update case.
   
   The new tests also pass on main, so the output matches the merge path. Ran 
the ci.yml steps locally (fmt, clippy, msrv, unit, rest-server, integration, 
datafusion, python, go).
   
   ### API and Format
   
   None.
   
   ### Documentation
   
   None.
   


-- 
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