Github user fhueske commented on the issue:
https://github.com/apache/flink/pull/4324
Thanks for the review @wuchong.
The `DataStream` join semantics can be implemented with this join by first
window grouping both inputs with a UDAGG that collects all records in a list,
and then joins the lists on the same time with `left.t BETWEEN right.t AND
right.t` (we do not support a single `==` predicate on time so we need `>=` and
`<=`, but this could be added, IMO). And then a UDF that crosses the lists.
I think this is how you would need to do it in SQL (you could use UNNEST to
do the Cartesian product). The Table API could have a syntactic shortcut for
that but the internal logical representation would again be the same.
Not sure if it adds much value. I think the windowed join semantics of this
PR are nicer because they do not have cut-off points at the window boundaries
(two records could be only 3 msecs apart but would not join because they are in
different tumbling windows).
---
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 enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
