Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-132585797
I also updated the documentation to reflect the changes in shipping
strategies/partitioning.
---
If your project is set up for it, you can reply to this email and have
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-132582717
What I did is basically adding every operator to the list of sinks. When
`execute` is called the translation starts from every operator, which is ok,
since an operator
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-132622278
I think this looks reasonable.
---
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
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-132689802
Manually merged
---
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
Github user aljoscha closed the pull request at:
https://github.com/apache/flink/pull/988
---
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
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-132260208
I think this is good now. I adapted the Streaming Tests to always use
`StreamingMultipleProgramTestBase` when appropriate. The earlier problems where
caused by some
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-131892889
I changed it to execute dangling operators now. There is, however, a
strange feature. This code works on master:
https://gist.github.com/aljoscha/bbe74309a31a16ca8413.
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-129563293
Sorry for the lack of activity. I'm currently on vacation and will pick
this up again when I'm back, next week.
---
If your project is set up for it, you can reply to
Github user StephanEwen commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-128405914
+1 for rebalancing automatically between operators of different DOP. The
batch API does the same. But it should really be rebalance, not a form of
forward that
Github user gyfora commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-128404999
It would be good to get some feedback from the others as well, but in
general my arguments are the following:
1. Getting exceptions after non-parallel sources
Github user aljoscha commented on a diff in the pull request:
https://github.com/apache/flink/pull/988#discussion_r36347244
--- Diff:
flink-staging/flink-streaming/flink-streaming-core/src/main/java/org/apache/flink/streaming/api/transformations/CoFeedbackTransformation.java
---
Github user aljoscha commented on a diff in the pull request:
https://github.com/apache/flink/pull/988#discussion_r36347361
--- Diff:
flink-staging/flink-streaming/flink-streaming-core/src/main/java/org/apache/flink/streaming/api/transformations/SinkTransformation.java
---
@@
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-128133833
About rebalance()/forward(). Yes, when the parallelism differs it throws an
exception now. Previously, when a user did not specify a partition strategy,
forward was
Github user gyfora commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-127924076
Why isn't rebalance implied when the 2 operators don't have the same
parallelism and partitioning is not defined? If you don't specify the
partitioning (which defaults to
Github user gyfora commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-127926607
If I understand correctly, this also this changes the semantics that we
execute programs without sinks, and also topology branches which don't end in
sinks. I personally
Github user mbalassi commented on a diff in the pull request:
https://github.com/apache/flink/pull/988#discussion_r36284249
--- Diff:
flink-staging/flink-streaming/flink-streaming-core/src/main/java/org/apache/flink/streaming/api/transformations/CoFeedbackTransformation.java
---
Github user mbalassi commented on a diff in the pull request:
https://github.com/apache/flink/pull/988#discussion_r36284613
--- Diff:
flink-staging/flink-streaming/flink-streaming-core/src/main/java/org/apache/flink/streaming/api/transformations/SinkTransformation.java
---
@@
Github user mbalassi commented on the pull request:
https://github.com/apache/flink/pull/988#issuecomment-127940968
I am not sure that I understand this correctly: If a non parallel source is
used does the user need to call `rebalance` to use all parallel instances of
the downstream
GitHub user aljoscha opened a pull request:
https://github.com/apache/flink/pull/988
[FLINK-2398][api-breaking] Introduce StreamGraphGenerator
This decouples the building of the StreamGraph from the API methods.
Before, the methods would build the StreamGraph as they go. Now the
19 matches
Mail list logo