[jira] (FLINK-5654) Add processing time OVER RANGE BETWEEN x PRECEDING aggregation to SQL

[radu (JIRA)]

Title: Message Title

radu commented on FLINK-5654   Re: Add processing time OVER RANGE BETWEEN x PRECEDING aggregation to SQL I would like to bring into the discussion two items: 1) For the query semantic I see that in the example the interval is strongly defined "RANGE BETWEEN INTERVAL '1' HOUR PRECEDING AND CURRENT ROW" I would propose to support also the syntax (used commonly in Calcite) "RANGE INTERVAL '1' HOUR PRECEDING" This syntax implicitly expects that the reference is the current row. 2)◾The PARTITION BY clause is optional (no partitioning results in single threaded execution). Is there a strong preference to use for the single tread execution one of the two options: -globalwindow (windowAll) - I would expect this to be used -keyBy(constantFunction).window().setParalelism(1) - this idea would keep things symmetric and with a unique implementation for the two cases (with partition and without). The disadvantage is that we create an addition operator with a dummy functionality (i.e., a keyBy with a constant function to ensure that all events arrive in the same place) to maintain the consistency in the implementation. Add Comment   This message was sent by Atlassian JIRA (v6.3.15#6346-sha1:dbc023d)