[
https://issues.apache.org/jira/browse/FLINK-6983?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16058959#comment-16058959
]
Dawid Wysakowicz edited comment on FLINK-6983 at 6/22/17 8:09 AM:
------------------------------------------------------------------
I agree that getting states by name would be the way to go, but it may happen
that after a change partial matches could be no longer valid for new graph of
states. Also the versioning of `SharedBuffer` is pretty fragile and after a
graph is changed there may be some paths collisions which can result in
incorrect matches.
For the second part, rather than changing an NFA, we would like to have
multiple NFAs, that may be injected in runtime. I think the new ValueState
would end up to be equal to NFA itself, cause we would need to move {{State}}
s, {{ComputationState}} s and {{SharedBuffer}} there ;)
was (Author: dawidwys):
I agree that getting states by name would be the way to go, but it may happen
that after a change partial matches could be no longer valid for new graph of
states. Also the versioning of `SharedBuffer` is pretty fragile and after a
graph is changed there may be some paths collisions which can result in
incorrect matches.
For the second part, rather than changing an NFA, we would like to have
multiple NFAs, that may be injected in runtime. I think the new ValueState
would end up to be equal to NFA itself, cause we would need to move {{State}}s,
{{ComputationState}}s and {{SharedBuffer}} there ;)
> Do not serialize States with NFA
> --------------------------------
>
> Key: FLINK-6983
> URL: https://issues.apache.org/jira/browse/FLINK-6983
> Project: Flink
> Issue Type: Improvement
> Components: CEP
> Reporter: Dawid Wysakowicz
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
