You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by kl0u <gi...@git.apache.org> on 2018/06/01 13:57:21 UTC
[GitHub] flink issue #6059: [Flink-9418] Migrate SharedBuffer to use MapState
Github user kl0u commented on the issue:
https://github.com/apache/flink/pull/6059
Hi @dawidwys! Thanks for the work.
The changes seem really good. As we discussed privately I did a review and the comments you can find them in the branch I sent you.
The only things I would like to write here as a reminder are:
1) as a future commit, we could in some cases do the cleaning up during discovering the partial matches, i.e. do the `releaseNode()` from within the `extractCurrentMatches()` (whenever appropriate). This could be a separate JIRA actually.
2) when migrating from an old version, we should ignore potential "dead-end" edges (edges that point to nodes that have been removed) in the partial match buffer.
---