[ 
https://issues.apache.org/jira/browse/LUCENE-5802?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14049633#comment-14049633
 ] 

David Smiley commented on LUCENE-5802:
--------------------------------------

The solution may be providing a FieldType in constructors to applicable 
Strategies.  There might be a utility class to facilitate generating queries 
(e.g. one that switches on the type borrowing similar code from Solr's 
TrieField) and other stuff.

> SpatialStrategy DocValues & FieldType customizability
> -----------------------------------------------------
>
>                 Key: LUCENE-5802
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5802
>             Project: Lucene - Core
>          Issue Type: Improvement
>          Components: modules/spatial
>            Reporter: David Smiley
>
> The SpatialStrategy API is a simple facade to index spatial data and query by 
> it in a consistent way, hiding the implementation.  For indexing data, it has 
> one applicable method:
> {code:java}
> public abstract Field[] createIndexableFields(Shape shape);
> {code}
> The base abstraction provides no further configuration. BBoxStrategy and 
> PointVectorStrategy have a way to set the precisionStep of the underlying 
> Double trie fields.  But none have a way to use Floats, and none have a way 
> to specify the use of DocValues (and which type).  Perhaps there are other 
> plausible nobs to turn.  It is actually more than just indexing since at 
> search time it may have to change accordingly (e.g. search difference between 
> Float & Double). PrefixTreeStrategy is likely to soon deprecate/remove any 
> applicability here (see LUCENE-5692).
> If there is no change that could reasonably be made to SpatialStrategy 
> itself, what is the pattern that BBoxStrategy and others should use?



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

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

Reply via email to