sunshone opened a new pull request, #13470:
URL: https://github.com/apache/gluten/pull/13470

   ## What changes are proposed in this pull request?
   
   Fixes #13469
   
   Maps the Flink temporal expressions to their velox flinksql implementations 
(companion velox PR bigo-sg/velox#93):
   
   - `RexCallConverterFactory`: `CURRENT_TIMESTAMP` / `NOW` / `LOCALTIMESTAMP` 
/ `LOCALTIME` entries — direct name mappings; these previously failed planning 
with "Function not supported".
   - `LogicalTypeConverter`: Flink RowData represents TIME as INT 
(millisecond-of-day) — the same encoding the velox flinksql `localtime` 
function uses — so `TimeType` maps onto velox INTEGER and TIME columns cross 
the bridge unchanged.
   
   The `localtimestamp` name exists only in the velox flinksql namespace 
(companion PR), so landing order / version pairing matters; `current_timestamp` 
/ `now` exist in the baseline and the companion PR registers a 
behavior-identical flinksql function under the same name (`current_timestamp`) 
plus the `now` alias, moving dialect ownership of the runtime path. 
`LOCALTIME`'s raw projection surfaces the carrier int where native Flink 
renders `HH:mm:ss` — an output-form divergence documented in the companion 
issue and ruled as tolerable by the downstream integration (the 
millisecond-of-day values are verified correct).
   
   ## How was this patch tested?
   
   - Planner-level: Maven `RexNodeConverterTest` 2/2; streaming operator suites 
(filter 8 run / 0 fail / 3 pre-existing skips; project 1/1).
   - End-to-end dual-run on Flink 1.19.2 (classpath-switched cluster, plan 
snapshots verify mode): 8 temporal SQL cases, all FINISHED on both engines — 
`localtimestamp` projections / filters / extract exact match after changelog 
folding; `current_timestamp` raw projections pass under decoupled criteria (row 
ids exact, wall-clock columns verified semantically on both engines); 
`localtime` cases pass under tolerance criteria (ids exact, carrier values 
verified in the `[0, 86'400'000)` domain); one GROUP BY case excluded — a 
pre-existing keyed-aggregation defect, reproduced identically before this 
change.


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to