[
https://issues.apache.org/jira/browse/FLINK-4098?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15346881#comment-15346881
]
Chesnay Schepler commented on FLINK-4098:
-----------------------------------------
At that point it is completely normal for the java side to wait. Under usual
circumstances the operator would be configured, read the broadcastVariables
(which java writes without waiting for any acknowledgement), and then
explicitly request data from java (until which java blocks).
So let's see. We know the python process is created, which is good. We also
know that the function is created, but never run. This means that something
went wrong with the setup of the operator, which reduces our area of failure to
the else block in Environment.execute(). I would (and will over the weekend :)
) plaster that stuff in debug statements.
Also, remember to call sys.stdout.flush() after print! This small detail has
caused me quite some trouble :)
> Iteration support in Python API
> -------------------------------
>
> Key: FLINK-4098
> URL: https://issues.apache.org/jira/browse/FLINK-4098
> Project: Flink
> Issue Type: New Feature
> Components: Python API
> Affects Versions: 1.0.2
> Reporter: Geoffrey Mon
> Priority: Minor
>
> Bulk and delta iterations are not supported in the Python API. Currently
> working on this at https://github.com/GEOFBOT/flink
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
