ryanschneider opened a new pull request, #452:
URL: https://github.com/apache/arrow-go/pull/452
Fixes #451.
### Rationale for this change
As mentioned in #451, the `RecordReader` returned from
`pqarrow.NewFileReader(...).GetRecordReader(...)` sets `RecordReader.Err()` to
`io.EOF` once all the records are read. This in turn causes
`flight.StreamChunksFromReader(...)` to propagate the `io.EOF` err in a
`StreamChunk{err: io.EOF}` at the end of the stream, leading to spurious errors.
### What changes are included in this PR?
At first I thought the fix was to change the implementation of
`recordReader.next(...)` however doing so lead to other issues, so I think the
cleanest change is to simply surpress returning `io.EOF` from
`recordReader.Err()` as in this PR.
### Are these changes tested?
Yes, I tested this changes w/ the server mentioned in #451 and by running
`PARQUET_TEST_DATA="$PWD/parquet-testing/data" go test ./...` with a recursive
clone of the repo.
### Are there any user-facing changes?
I guess this could be considered user-facing since the behavior of the
returned `RecordReader.Err()` is changing, it's possible consumers are
expecting this to return `io.EOF` at the end.
--
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: github-unsubscr...@arrow.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
