Github user uce commented on the issue:
https://github.com/apache/flink/pull/3442
I had a quick chat with Stephan about this. @StefanRRichter has an idea how
to properly implement this. Closing this PR and unassigning the issue.
---
Github user zentol commented on the issue:
https://github.com/apache/flink/pull/3442
What's the status of this PR @uce @tillrohrmann @StefanRRichter
@StephanEwen ?
---
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
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/3442
Are we gonna get this in for 1.3?
---
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
Github user StefanRRichter commented on the issue:
https://github.com/apache/flink/pull/3442
Very much ok with me, my vote on this was actually on keeping old versioned
code duplicated and immutable.
---
If your project is set up for it, you can reply to this email and have your
Github user uce commented on the issue:
https://github.com/apache/flink/pull/3442
I agree with the equals/hashCode question. It has been introduced for the
various involved classes by different people and not in this PR, so I think
it's best handled as part of a different issue.
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/3442
Another thought from the discussion with @StefanRRichter :
You refactor a lot to not have duplicate code. While this is good in
general, I am wondering if we should not actually duplicate the
Github user StephanEwen commented on the issue:
https://github.com/apache/flink/pull/3442
Looks good in general.
One thing that I stumbled across a lot in recent works on the checkpoints /
savepoints is that they all implement `equals` and `hashCode` and delegate to
the task
Github user uce commented on the issue:
https://github.com/apache/flink/pull/3442
Thanks for your review Stefan! I addressed your comments, but only then
realized that the restriction to relative file state handle is actually a
problem for externalized checkpoints. :-( They possibly