Github user zentol commented on the issue:
https://github.com/apache/flink/pull/4485
merging.
---
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , I missed that message of `verifyAllBuffersReturned()` issue before.
I have submitted the modifications of it. :)
---
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , sorry for the typos. I have submitted the updates.
---
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , I have submitted the updates.
---
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , thank you for always helpful reviews!
I am very busy improving the runtime for our singles day these days. I will
submit the updates for this PR later today. And I also plan to
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , I have submitted the updates based on last comments.
Your concerns of interaction between requesting memory segments and
creating buffer pool is really necessary. I also noticed
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , thank you for so detail and helpful comments!
I will be on team outing tomorrow and come back on Sunday. I would consider
you concerns carefully and may submit the updates next
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , I have submitted the updates based on the above comments. :(
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
Yes, this way also has some advantages, and recycling these exclusive
buffers would be covered in next PR with some additional tests.
I will consider your suggestions to supplement some
Github user NicoK commented on the issue:
https://github.com/apache/flink/pull/4485
Hi @zhijiangW,
regarding the buffer pool implementation, I was just curious about why it
was done that way. But it is fine to keep the logic in `RemoteInputChannel` if
you make sure, that a
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
For `ResultPartitionType` comment, I expand to add the 'isCreditBased'
field temporarily in order not to affect the current process. My initial idea
is to remove this field after the whole feature
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , thank you for giving this discussion and comments!
Actually I proposed the same way of using fixed-size `LocalBufferPool` for
managing exclusive buffers for per
Github user NicoK commented on the issue:
https://github.com/apache/flink/pull/4485
Reviewed 9 of 13 files at r2.
Review status: all files reviewed at latest revision, 7 unresolved
discussions, some commit checks failed.
---
*[a
Github user zhijiangW commented on the issue:
https://github.com/apache/flink/pull/4485
@NicoK , I have submitted the updates:
- Create the fix size `LocalBufferPool` for floating buffers
- Assign the exclusive buffers for `InputChannel` directly
- The proposed
14 matches
Mail list logo