zenfenan commented on issue #11474:
URL: https://github.com/apache/flink/pull/11474#issuecomment-618300290
Understood. I assume this suggestion comes from the belief that the users
may update or add new metadata on the fly based on some external factor (either
based on the input element
zenfenan commented on issue #11474:
URL: https://github.com/apache/flink/pull/11474#issuecomment-616701185
@kl0u I agree with the point that the VectorizedRowBatch shouldn't be
exposed to the implementor. I am changing `vectorize()` method to take
VectorizedRowBatch as an argument along
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-615831483
@kl0u I have moved the implementation to `flink-orc` module. I think it
should be easier to take a complete look now. :)
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-615329178
@kl0u I checked your branch and I think it looks exactly like what you had
commented earlier. My changes pretty much are
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-614053436
Yes, @kl0u
I am moving the codebase to `flink-orc` but before that I think we need to
bump ORC dependency
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-612987289
@kl0u Not at all. In fact, I have added `appendUserMetadata()` already in
the latest commit which I pushed yesterday. As
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-61261
@kl0u @JingsongLi
I have updated the PR with the changes. Appreciate if you could take a look.
Thanks :)
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-612393911
@kl0u On the Vectorizer improvement, I believe we cannot have the
addUserMetadata() in Vectorizer class because it needs
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-611569646
@kl0u Hey Kostas. Thanks for the suggestions. I agree with the improvements
in `Vectorizer`. I actually had already made
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-607620461
@kl0u and @JingsongLi Thank you so much for reviewing the work. Regarding
the the way `VectorizedRowBatch` gets handled,
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-605911657
Cool. I'll update the PR.
This is an automated message
zenfenan commented on issue #11474: FLINK-10114: Add ORC BulkWriter support for
StreamingFileSink
URL: https://github.com/apache/flink/pull/11474#issuecomment-605903600
@kl0u Got it. I can do that. Where should that go in? Is it okay if I just
add them in Javadoc comments itself?
12 matches
Mail list logo