[
https://issues.apache.org/jira/browse/FLINK-5823?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16321900#comment-16321900
]
ASF GitHub Bot commented on FLINK-5823:
---------------------------------------
Github user StefanRRichter commented on the issue:
https://github.com/apache/flink/pull/5248
Yes, that is a pretty good summary. `CompletedCheckpointsRegistry` or maybe
just `CompletedCheckpoints` might already be a good enough name, after all it
is just a queue.
One more comment about the typed pointer, I think there is a small
misunderstanding. I don't ask to type it for the content, but for the usage. I
agree that this should just a a string provided by the user, but wrapping it
until final usage (i.e. interpretation) has advantages:
- If you see the variable, you know what it stands for and it requires no
further thinking about what that object could be - you see it immediately or by
hovering over it.
- Functions like xyz(String, String, String) are strictly worse than
xyz(JobID, OperatorName, StoragePointer), you need to think more and can make
more mistakes be messing up parameter order. Similar to why we have all the id
types and not just use `UUID` everywhere.
- If you search usage of that type, you can easily track all code that
deals with storage pointers.
- This type is rarely produced in the system, so memory overhead should not
count more than the above points.
> Store Checkpoint Root Metadata in StateBackend (not in HA custom store)
> -----------------------------------------------------------------------
>
> Key: FLINK-5823
> URL: https://issues.apache.org/jira/browse/FLINK-5823
> Project: Flink
> Issue Type: Sub-task
> Components: State Backends, Checkpointing
> Reporter: Stephan Ewen
> Assignee: Stephan Ewen
> Priority: Blocker
> Fix For: 1.5.0
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
