hcrosse opened a new pull request, #1606:
URL: https://github.com/apache/arrow-go/pull/1606

   ### Rationale for this change
   
   `ReadBatch` and `ReadBatchInPage` return level and value decoding errors 
without recording them, so `HasNext` stays true. On optional columns a retry 
consumes the next definition levels and drops values. After a failed 
`SeekToRow`, later reads continue from the following page, and the record 
reader decodes levels that it buffered before the seek.
   
   ### What changes are included in this PR?
   
   The column reader stores decode and seek errors, including the error for a 
page that holds fewer values than its header claims. The record reader clears 
buffered levels when a seek or page reader replacement fails.
   
   ### Are these changes tested?
   
   I added `TestReadBatchStopsAtDecodeError`, `TestReadBatchStopsAtNoProgress`, 
`TestSeekToRowToCorruptPageStopsReader`, 
`TestRecordReaderStopsAfterPageReaderFailure` and 
`TestRecordReaderSeekToCorruptPage`. All fail on `main` and pass with this 
change.
   
   I ran `go test -race -tags assert,test ./parquet/file ./parquet/pqarrow` and 
`golangci-lint` 2.13.2 with Go 1.26.10 on macOS arm64. Both passed.
   
   ### Are there any user-facing changes?
   
   Yes. After a read or seek error, including a `SeekToRow` to a row out of 
range, reads return that error until a seek succeeds. There is no API 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]

Reply via email to