[
https://issues.apache.org/jira/browse/BEAM-2140?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16065398#comment-16065398
]
Aljoscha Krettek commented on BEAM-2140:
----------------------------------------
I need to look into things a bit but I think your comment about needing a
"loop" is right: at the end of the {{@ProcessElement}} method a processing-time
timer is set for the current processing time. This is expected to "re-awake"
the {{@ProcessElement}} method, thereby creating a loop that runs until we have
no more residual restriction to process. (Or, as is the case now, until the
watermark on the input goes to +Inf, thereby missing some processing.)
As a side node, the timers that {{SplittableDoFnOperator}} currently hands to
{{ProcessFn}} (via {{StatefulDoFnRunner}}, which does late data dropping, i.e.
also timer dropping) are wrapped in a {{WindowedValue}} with a global window
and this global window is used for determining if the timer is late, which is
only the case if the watermark went to +Inf. This is more by chance than by any
conscious decision.
> Fix SplittableDoFn ValidatesRunner tests in FlinkRunner
> -------------------------------------------------------
>
> Key: BEAM-2140
> URL: https://issues.apache.org/jira/browse/BEAM-2140
> Project: Beam
> Issue Type: Bug
> Components: runner-flink
> Reporter: Aljoscha Krettek
> Assignee: Aljoscha Krettek
>
> As discovered as part of BEAM-1763, there is a failing SDF test. We disabled
> the tests to unblock the open PR for BEAM-1763.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
