[
https://issues.apache.org/jira/browse/LUCENE-7899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16074744#comment-16074744
]
Uwe Schindler commented on LUCENE-7899:
---------------------------------------
bq. ElasticSearch indexes its own field to record which field names occur in a
document, so it's able to do "exists" for any field (not just doc values
fields), but I think doc values fields we can just get "for free".
IMHO, this is still the preferable method of doing this as you only need one
field and you can quickly lookup all documents with a simple inverted index
query. I generally recommend the same strategy also to Solr users (they just
have to do it manually). Index size is in most cases not a problem, as the term
index is small and the posting list is highly compressed!
> Rename FieldValueQuery to DocValuesFieldExistsQuery
> ---------------------------------------------------
>
> Key: LUCENE-7899
> URL: https://issues.apache.org/jira/browse/LUCENE-7899
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Michael McCandless
> Assignee: Michael McCandless
> Priority: Blocker
> Fix For: 7.0
>
> Attachments: LUCENE-7899.patch
>
>
> I don't think we have a query today to efficiently test whether a doc values
> field exists (has any value) for each document in the index?
> Now that we use iterators to access doc values, this should be an efficient
> query: we can return the DISI we get for the doc values.
> ElasticSearch indexes its own field to record which field names occur in a
> document, so it's able to do "exists" for any field (not just doc values
> fields), but I think doc values fields we can just get "for free".
> I haven't started on this ... just wanted to open the issue first for
> discussion.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
