[
https://issues.apache.org/jira/browse/LUCENE-1476?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12665965#action_12665965
]
Michael McCandless commented on LUCENE-1476:
--------------------------------------------
bq. I am going to run the contrib benchmark tests on this patch vs trunk to see
if there is a difference in performance
I think this is the most important next step before we go too much
further.
There are two separate questions we are exploring here:
1) Should we switch to iterator only API for accessing deleted docs
(vs random-access API we use today)?
2) Should we take deletes into account at a higher level
(BooleanScorer/2, as a top-level Filter) vs the lowest level
(each posting, in SegmentTermDocs that we do today)?
We need to understand the performance impact with each combination of
1) and 2), across different queries. Likely the answer will be "it
depends on the query", but we should understand just how much. I
think performance cost/benefit is the driver here,
> BitVector implement DocIdSet, IndexReader returns DocIdSet deleted docs
> -----------------------------------------------------------------------
>
> Key: LUCENE-1476
> URL: https://issues.apache.org/jira/browse/LUCENE-1476
> Project: Lucene - Java
> Issue Type: Improvement
> Components: Index
> Affects Versions: 2.4
> Reporter: Jason Rutherglen
> Priority: Trivial
> Attachments: LUCENE-1476.patch, LUCENE-1476.patch, LUCENE-1476.patch,
> quasi_iterator_deletions.diff, quasi_iterator_deletions_r2.diff
>
> Original Estimate: 12h
> Remaining Estimate: 12h
>
> Update BitVector to implement DocIdSet. Expose deleted docs DocIdSet from
> IndexReader.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-dev-h...@lucene.apache.org
