Github user rtudoran commented on the issue:
https://github.com/apache/flink/pull/3783
@fhueske @stefanobortoli
Regarding the options for solving the distinct. From my point of view the
previous approach worked:
- we keep in the processFunctions a state for each field that is used in a
distinct aggregator
- we count the occurrence of each value (meant for a distinct aggregate)
that we observed
- when a value is seen for the first time we accumulate it
- when a value is retracted we decrease the corresponding count.
-if count is 0 we retract the value from accumulator
Based on how things are implemented now - this would involved to have a
separate list of aggregatefunctions for the distinct. In order to be able to
control when to accumulate to these values.
What do you think? Do you see any disadvantage to this?
---
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.
---
