[jira] [Updated] (LUCENE-5429) Run one search across multiple scorers/collectors

2014-03-12 Thread Michael McCandless (JIRA)

 [ 
https://issues.apache.org/jira/browse/LUCENE-5429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Michael McCandless updated LUCENE-5429:
---

Attachment: LUCENE-5429.patch

This is a patch against 4.3.1 (small changes were required because SimScorer 
API changes); it also includes the QueryRescorer (LUCENE-5489).

> Run one search across multiple scorers/collectors
> -
>
> Key: LUCENE-5429
> URL: https://issues.apache.org/jira/browse/LUCENE-5429
> Project: Lucene - Core
>  Issue Type: Improvement
>Reporter: Michael McCandless
>Assignee: Michael McCandless
> Attachments: LUCENE-5429.patch, LUCENE-5429.patch, LUCENE-5429.patch
>
>
> I'm looking into the possibility of running the same search across many 
> scorers, so that decoding postings lists / doing union and intersect are done 
> once, but scoring via Similarity can be done multiple times for each it (and 
> the results collected into separate collectors).



--
This message was sent by Atlassian JIRA
(v6.2#6252)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Updated] (LUCENE-5429) Run one search across multiple scorers/collectors

2014-03-05 Thread Michael McCandless (JIRA)

 [ 
https://issues.apache.org/jira/browse/LUCENE-5429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Michael McCandless updated LUCENE-5429:
---

Attachment: LUCENE-5429.patch

Here's a new patch, fixing DisjunctionSum/MaxScorer to not score during 
matching.

I sort of think this is something we should separately explore ... e.g. if you 
do a Filter with one of these queries, they are scoring unnecessarily.  Queries 
really shouldn't score during matching.

For this issue, the test now passes, so you can run matching once and scoring N 
times to N collectors, for disjunctions.

> Run one search across multiple scorers/collectors
> -
>
> Key: LUCENE-5429
> URL: https://issues.apache.org/jira/browse/LUCENE-5429
> Project: Lucene - Core
>  Issue Type: Improvement
>Reporter: Michael McCandless
>Assignee: Michael McCandless
> Attachments: LUCENE-5429.patch, LUCENE-5429.patch
>
>
> I'm looking into the possibility of running the same search across many 
> scorers, so that decoding postings lists / doing union and intersect are done 
> once, but scoring via Similarity can be done multiple times for each it (and 
> the results collected into separate collectors).



--
This message was sent by Atlassian JIRA
(v6.2#6252)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org



[jira] [Updated] (LUCENE-5429) Run one search across multiple scorers/collectors

2014-02-01 Thread Michael McCandless (JIRA)

 [ 
https://issues.apache.org/jira/browse/LUCENE-5429?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Michael McCandless updated LUCENE-5429:
---

Attachment: LUCENE-5429.patch

Patch, making a Similarity that switches between sub-Sims, and a custom 
collector that uses that switching and calls .score() and then collects into 
multiple sub-Collectors.

While this works for TermQuery, and all-MUST BooleanQuery, the test fails on 
all-SHOULD BooleanQuery because DisjunctionSum/MaxScorer pre-compute the score 
during matching.  I know they do this for performance reasons (single pass 
recursion through the PQ to find the sub-scorers that matched)... so I think 
the only way this feature can work in general is with a custom BQ that doesn't 
do this opto.

> Run one search across multiple scorers/collectors
> -
>
> Key: LUCENE-5429
> URL: https://issues.apache.org/jira/browse/LUCENE-5429
> Project: Lucene - Core
>  Issue Type: Improvement
>Reporter: Michael McCandless
>Assignee: Michael McCandless
> Attachments: LUCENE-5429.patch
>
>
> I'm looking into the possibility of running the same search across many 
> scorers, so that decoding postings lists / doing union and intersect are done 
> once, but scoring via Similarity can be done multiple times for each it (and 
> the results collected into separate collectors).



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

-
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org