[
https://issues.apache.org/jira/browse/FLINK-1092?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14124540#comment-14124540
]
Fabian Hueske commented on FLINK-1092:
--------------------------------------
I think the reason for this restriction is that the solution set is internally
stored in a distributed hash table.
In regular (non-iterative) processing, Flink uses hash tables only for join
operations wich access a hash table by probing it. Other operations like Map or
Filter would require to read the full hash table and my guess is that this has
not been implemented because there was no use case for that so far. Support for
cogroup was probably added because it was required for some algorithm.
[~StephanEwen] Is that correct?
> Allow operations, other than join and cogroup, on the solution set
> ------------------------------------------------------------------
>
> Key: FLINK-1092
> URL: https://issues.apache.org/jira/browse/FLINK-1092
> Project: Flink
> Issue Type: Improvement
> Components: Iterations
> Reporter: Vasia Kalavri
> Priority: Minor
> Labels: feature
>
> Currently, join and cogroup are the only operations allowed on the solution
> set. We should make the solution set usable for any type of operation,
> possibly using special functions.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
