You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@flink.apache.org by "Jun Qin (Jira)" <ji...@apache.org> on 2020/02/03 12:30:00 UTC
[jira] [Created] (FLINK-15865) When to add .uid() call:
inconsistent definition of operators in Flink docs
Jun Qin created FLINK-15865:
-------------------------------
Summary: When to add .uid() call: inconsistent definition of operators in Flink docs
Key: FLINK-15865
URL: https://issues.apache.org/jira/browse/FLINK-15865
Project: Flink
Issue Type: Bug
Components: Documentation
Affects Versions: 1.9.1
Reporter: Jun Qin
On one hand, the Flink doc suggests to add .uid() call for *all* operators in [1], on the other hand, it lists all operators in Flink [2]. The issues are:
# KeyBy is listed as an operator, but .keyBy().uid() is not a valid call. This is same for window(), split(), etc.
# addSource(), addSink() are not listed as operators, but we do expect user to call .uid() after addSource() and addSink(), especially in the exact-once scenario.
This creates confusions, esp., for beginners. There should be a better definition about which/what kind of operators can have a following uid() call.
[1] [Should I assign ids to all operators in my job|https://ci.apache.org/projects/flink/flink-docs-release-1.9/ops/state/savepoints.html#should-i-assign-ids-to-all-operators-in-my-job]
[2] [Flink Operators|https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/stream/operators/]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)