[ 
https://issues.apache.org/jira/browse/LUCENE-5454?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13905456#comment-13905456
 ] 

Adrien Grand commented on LUCENE-5454:
--------------------------------------

+1 I like the approach and the fact that this is an optional API

> SortField for SortedSetDV
> -------------------------
>
>                 Key: LUCENE-5454
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5454
>             Project: Lucene - Core
>          Issue Type: New Feature
>          Components: modules/other
>            Reporter: Robert Muir
>         Attachments: LUCENE-5454.patch
>
>
> Currently, its not possible to sort by a sortedsetdv (e.g. a multi-valued 
> field). So the idea is to provide some comparators that let you do this: by 
> choosing a selector (e.g. min/max/median type stuff) that picks a value from 
> the set as a representative sort value.
> The implementation is pretty simple, just actually wrap the SortedSet in a 
> SortedDV that does the selection, and re-use the existing 
> TermOrdValComparator logic.
> One issue is that, with the current sortedset API only 'min' is a viable 
> option because its the only one that can be done in constant time. So this 
> patch adds an optional extension (RandomAccessOrds) for codecs that can 
> support random access. I added this to the default codec, diskdv, and direct, 
> as they can all easily support it. While this could be useful for other 
> purposes (e.g. min/max as valuesource or whatever), i think its best to be 
> optional because it prevents some forms of encoding/compression.
> Anyway I'm targeting lucene/sandbox with this change... 



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to