[
https://issues.apache.org/jira/browse/COUCHDB-389?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12721509#action_12721509
]
Brad Anderson commented on COUCHDB-389:
---------------------------------------
as davisp pointed out in IRC, this patch would do a reduce for every row...
sub-optimal.
Also, I've reverted this patch, and the clustering code is still working. I
believe a change elsewhere fixed my issue (using Total Count when no values
were returned from the shard). I'll do some more checks, and if that's the
case, I'll close this as invalid.
> [PATCH] Make offset calculation more available during btree folding
> -------------------------------------------------------------------
>
> Key: COUCHDB-389
> URL: https://issues.apache.org/jira/browse/COUCHDB-389
> Project: CouchDB
> Issue Type: Improvement
> Components: Database Core
> Affects Versions: 0.9
> Reporter: Brad Anderson
> Priority: Minor
> Attachments: clustered_offset.patch
>
>
> When CouchDB returns no values for _all_docs (or potentially any fold over
> the btree), then no offset is calculated, either. This is not a problem in
> non-clustered CouchDB. However, when you are in a clustered environment and
> treat each shard as its own CouchDB, then not returning an offset when no
> values are found on the shard can be problematic. The specific case is if
> there is a startkey or skip specified, and there were values before (or after
> with descending=true) the startkey or skip values, but there are no current
> values.
> Moving the Offset calculation to be available for all clauses in the return
> fun of couch_httpd_view:make_view_fold_fun/6 allows the user to store Offsets
> in the RowFunAcc and client code using this function will work as expected.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
