[
https://issues.apache.org/jira/browse/LUCENE-1483?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michael McCandless updated LUCENE-1483:
---------------------------------------
Attachment: LUCENE-1483.patch
New patch attached:
* I moved maxScore tracking up into TopFieldCollector to save a
method call per collect().
* Deprecated TopDocs.get/setMaxScore()
* Inlined the common "only 1 comparator" case
* Some small optimizations
* Fixed a bug in reverse sorting & short-circuit testing
* Renamed a few attrs
In addition to fixing TestSort to test multi-segment index where
totalHits is much greater than topN, we should include reverse sorting
(to tickle the bug I just found) as well as sorting by 2 or more
fields. Mark do you want to take a stab at this? Given how many
sneaky bugs we're uncovering just by staring at the code (for long
enough) I'd like to increase test coverage....
Here're the topN=10 results with this patch:
||numSeg||index||sortBy||method||query||topN||hits||warm||qps||warmnew||qpsnew||pctg||
|1|simple|country|val|text|10|2000000| 0.6| 10.7| 0.7| 11.9| 11.2%|
|1|simple|country|ord|text|10|2000000| 0.6| 10.7| 0.6| 14.2| 32.7%|
|1|simple|country|ordval|text|10|2000000| 0.6| 10.7| 0.6| 14.3| 33.6%|
|1|simple|country|orddem|text|10|2000000| 0.6| 10.7| 0.6| 13.4| 25.2%|
|1|wiki|title|val|147|10| 4984| 2.1|3743.8| 2.4|2451.8|-34.5%|
|1|wiki|title|ord|147|10| 4984| 2.1|3743.8| 2.1|4459.4| 19.1%|
|1|wiki|title|ordval|147|10| 4984| 2.1|3743.8| 2.1|4478.2| 19.6%|
|1|wiki|title|orddem|147|10| 4984| 2.1|3743.8| 2.0|4233.9| 13.1%|
|1|wiki|title|val|text|10| 97191| 2.1| 144.2| 2.4| 38.9|-73.0%|
|1|wiki|title|ord|text|10| 97191| 2.1| 144.2| 2.1| 165.0| 14.4%|
|1|wiki|title|ordval|text|10| 97191| 2.1| 144.2| 2.1| 159.9| 10.9%|
|1|wiki|title|orddem|text|10| 97191| 2.1| 144.2| 2.1| 161.4| 11.9%|
|1|wiki|title|val|1|10| 386435| 2.1| 51.2| 2.4| 13.5|-73.6%|
|1|wiki|title|ord|1|10| 386435| 2.1| 51.2| 2.1| 67.1| 31.1%|
|1|wiki|title|ordval|1|10| 386435| 2.1| 51.2| 2.1| 66.6| 30.1%|
|1|wiki|title|orddem|1|10| 386435| 2.1| 51.2| 2.1| 64.7| 26.4%|
||numSeg||index||sortBy||method||query||topN||hits||warm||qps||warmnew||qpsnew||pctg||
|10|simple|country|val|text|10|2000000| 0.7| 10.4| 0.7| 11.6| 11.5%|
|10|simple|country|ord|text|10|2000000| 0.7| 10.4| 0.5| 13.9| 33.7%|
|10|simple|country|ordval|text|10|2000000| 0.7| 10.4| 0.5| 14.0| 34.6%|
|10|simple|country|orddem|text|10|2000000| 0.7| 10.4| 0.5| 13.2| 26.9%|
|10|wiki|title|val|147|10| 4984| 12.5|3004.5| 2.6|1695.3|-43.6%|
|10|wiki|title|ord|147|10| 4984| 12.5|3004.5| 2.1|3072.8| 2.3%|
|10|wiki|title|ordval|147|10| 4984| 12.5|3004.5| 2.1|3328.7| 10.8%|
|10|wiki|title|orddem|147|10| 4984| 12.5|3004.5| 2.1|3295.1| 9.7%|
|10|wiki|title|val|text|10| 97191| 12.7| 139.4| 2.4| 38.7|-72.2%|
|10|wiki|title|ord|text|10| 97191| 12.7| 139.4| 2.1| 158.9| 14.0%|
|10|wiki|title|ordval|text|10| 97191| 12.7| 139.4| 2.1| 161.7| 16.0%|
|10|wiki|title|orddem|text|10| 97191| 12.7| 139.4| 2.1| 157.7| 13.1%|
|10|wiki|title|val|1|10| 386435| 12.7| 50.3| 2.5| 15.6|-69.0%|
|10|wiki|title|ord|1|10| 386435| 12.7| 50.3| 2.1| 65.4| 30.0%|
|10|wiki|title|ordval|1|10| 386435| 12.7| 50.3| 2.1| 66.4| 32.0%|
|10|wiki|title|orddem|1|10| 386435| 12.7| 50.3| 2.1| 63.5| 26.2%|
||numSeg||index||sortBy||method||query||topN||hits||warm||qps||warmnew||qpsnew||pctg||
|100|simple|country|val|text|10|2000000| 1.0| 8.8| 3.1| 9.5| 8.0%|
|100|simple|country|ord|text|10|2000000| 1.0| 8.8| 0.6| 11.4| 29.5%|
|100|simple|country|ordval|text|10|2000000| 1.0| 8.8| 0.6| 11.3| 28.4%|
|100|simple|country|orddem|text|10|2000000| 1.0| 8.8| 0.6| 11.0| 25.0%|
|100|wiki|title|val|147|10| 4984| 94.6|1066.9| 3.7| 456.8|-57.2%|
|100|wiki|title|ord|147|10| 4984| 94.6|1066.9| 2.1| 522.2|-51.1%|
|100|wiki|title|ordval|147|10| 4984| 94.6|1066.9| 2.1| 667.1|-37.5%|
|100|wiki|title|orddem|147|10| 4984| 94.6|1066.9| 2.1| 781.7|-26.7%|
|100|wiki|title|val|text|10| 97191| 94.9| 110.2| 2.8| 38.4|-65.2%|
|100|wiki|title|ord|text|10| 97191| 94.9| 110.2| 2.1| 123.0| 11.6%|
|100|wiki|title|ordval|text|10| 97191| 94.9| 110.2| 2.2| 127.3| 15.5%|
|100|wiki|title|orddem|text|10| 97191| 94.9| 110.2| 2.2| 126.8| 15.1%|
|100|wiki|title|val|1|10| 386435| 94.3| 47.9| 2.8| 15.8|-67.0%|
|100|wiki|title|ord|1|10| 386435| 94.3| 47.9| 2.1| 59.8| 24.8%|
|100|wiki|title|ordval|1|10| 386435| 94.3| 47.9| 2.2| 60.8| 26.9%|
|100|wiki|title|orddem|1|10| 386435| 94.3| 47.9| 2.2| 59.0| 23.2%|
> Change IndexSearcher to use MultiSearcher semantics for multiple subreaders
> ---------------------------------------------------------------------------
>
> Key: LUCENE-1483
> URL: https://issues.apache.org/jira/browse/LUCENE-1483
> Project: Lucene - Java
> Issue Type: Improvement
> Affects Versions: 2.9
> Reporter: Mark Miller
> Priority: Minor
> Attachments: LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch,
> LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch,
> LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch,
> LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch,
> LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch, LUCENE-1483.patch,
> sortBench.py, sortCollate.py
>
>
> FieldCache and Filters are forced down to a single segment reader, allowing
> for individual segment reloading on reopen.
--
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: [email protected]
For additional commands, e-mail: [email protected]
