[
https://issues.apache.org/jira/browse/OAK-5878?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15899217#comment-15899217
]
Marcel Reutegger commented on OAK-5878:
---------------------------------------
bq. Actually, once the split document has been disconnected, and could be
deleted right away, no?
Yes, this will work. I think the reason why the current implementation has
those two distinct phases is the optimization for the MongoDB case. As you
mentioned already that implementation removes the split documents at the end
with a single call.
> SplitDocumentCleanup iterates twice over splitDocGarbage
> --------------------------------------------------------
>
> Key: OAK-5878
> URL: https://issues.apache.org/jira/browse/OAK-5878
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: documentmk
> Reporter: Julian Reschke
> Assignee: Julian Reschke
> Attachments: OAK-5878.diff
>
>
> {{SplitDocumentCleanup}} currently iterates twice over {{splitDocGarbage}}.
> NOTE: not the case for Mongo DB, as {{MongoVersionGCSupport}} overwrites
> {{deleteSplitDocuments()}}.
> {{deleteSplitDocuments()}} currently iterates over {{splitDocGarbage}} to
> obtain the IDs of the documents to be deleted. Instead, we could just collect
> the IDs inside {{disconnect()}}, the memory requirements would be the same.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
