vbhanuchander-lang commented on PR #28780:
URL: https://github.com/apache/flink/pull/28780#issuecomment-5526709132

   @nikitasinghvi this looks right to me, and it reads as an omission rather 
than a decision.
   
   `FlinkParquetBuilder.createWriter` already wires seven conf values through 
the same
   `withX(getX(conf))` shape — `withRowGroupSize(getBlockSize(...))`, 
`withPageSize`,
   `withDictionaryPageSize`, `withDictionaryEncoding`, `withValidation`, 
`withWriterVersion`,
   `withCompressionCodec`. These two are the only `ParquetOutputFormat.get*` 
accessors of that family
   missing from the chain, so the change follows the file's own convention 
rather than introducing one.
   
   Checked what I could without building:
   
   - `master`'s `ParquetRowDataBuilder` contains no reference to 
`withMinRowCountForPageSizeCheck`,
     `withMaxRowCountForPageSizeCheck`, or the two conf keys, so those keys are 
genuinely ignored today.
   - Both accessors and both builder methods exist with the signatures used 
here —
     `ParquetOutputFormat.getMinRowCountForPageSizeCheck(Configuration)` /
     `getMaxRowCountForPageSizeCheck(Configuration)` returning `int`, and
     `ParquetWriter.Builder.withMinRowCountForPageSizeCheck(int)` /
     `withMaxRowCountForPageSizeCheck(int)` (verified against 
`parquet-hadoop-1.16.0`).
   
   **One thing I could not settle and would want a reviewer to confirm:** the 
`.withConf(conf)` call
   sits *after* the two new calls in the chain. For the seven existing values 
that ordering is
   evidently fine, so it very likely is here too, but it is the one way this 
change could silently do
   nothing — worth a reviewer's eye or an assertion in the test that the 
configured value actually
   reaches the writer, rather than only that the write succeeds.
   
   @MartijnVisser you merged the most recent `flink-parquet` change (#28565) — 
would you be willing to
   take a look? This has been sitting since 28 July with no review, and it is 
+52/-0 with a test.
   


-- 
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