Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-191683975
Okay, that sounds like a design flaw in hiding the aggregating functions
completely from the window operator in the state. I think the operator should
be aware of th
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-191681857
It is certainly possible to find some way of doing it but it is not
straightforward. So for now I wanted to keep it simple until we figure out a
good way to do it.
Github user wenlonglwl commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-191081091
3qï¼ @aljoscha ~ I agree with @StephanEwen, your purpose is to prevent
user from using partitioned state in windowed reduce function, but removing
rich functions do
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-190769526
I am wondering whether we cannot allow rich functions in the windows still.
The window operator would need to call `open()`, and `close()`rather than the
state. What
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-190649095
Hi @wenlonglwl, the reason for this were mostly practical concerns. We want
to use our partitioned state abstraction for the window state because that
makes it easy to
Github user wenlonglwl commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-190511772
Hi, @aljoscha , I am confused about the change you make in this issue,
that makes RichFunction not supported in the aggregation function of the
windowed stream, wha
Github user aljoscha closed the pull request at:
https://github.com/apache/flink/pull/1562
---
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 project does not have this feature
enabled and wishes so, or if the feature is en
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-178195987
Rebased and extended this pull request in #1571
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If yo
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-178146614
Have rebased this onto the current master.
One thing I'd like to change is to move the HDFS copy process utils from
`flink-core` to `flink-runtime`.
Idea
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-178130142
Okay, nice, didn't look at the code after the changes again.
+1 to merge this then
---
If your project is set up for it, you can reply to this email and hav
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-178111578
@StephanEwen @gyfora I have the changes reflected in this PR already.
---
If your project is set up for it, you can reply to this email and have your
reply appear on Gi
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-178099623
I think this is in quite good shape.
@gyfora's comment makes sense, +1 for such a default implementation.
The changes I am making are based on this,
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-177839471
@gyfora you're right, I will add default implementations for ListState and
ReducingState and use them in the DbStateBackend. For RocksDB there are custom
implementation
Github user gyfora commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-177611887
This would also remove a lot of code duplication for the DbStateBackend and
would probably also make the RocksDb backend a little cleaner :)
---
If your project is set u
Github user gyfora commented on the pull request:
https://github.com/apache/flink/pull/1562#issuecomment-177611308
Hey,
First of all, great work I am looking forward to having this is :)
I think it would be good if we added default implementations of the List,
Reducing sta
GitHub user aljoscha opened a pull request:
https://github.com/apache/flink/pull/1562
Enhance Partitioned State and use it in WindowOperator
The commits in this are self-contained. The first one enhances the
partitioned state such that it can be used by the WindowOperator. The secon
16 matches
Mail list logo
