asfimport

Adrien Grand (@jpountz) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17484830&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17484830)) One interesting question about this will be the read API. What I have in mind for now looks something like this: ```java class SparseIndexProducer...

Robert Muir (@rmuir) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17485178&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17485178)) an entire new file format abstraction for this one slow range query? is there any other use-case for this data? I'm concerned about complexity,...

Adrien Grand (@jpountz) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17485229&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17485229)) Actually I think it would help make things faster not slower. IndexSortSortedNumericDocValuesRangeQuery already outperforms PointRangeQuery for ranges that match lots of documents, even more...

Robert Muir (@rmuir) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17485292&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17485292)) thanks @jpountz. I like that there is a sort use-case as well, although it is still an advanced/complex case (multi-field sort). What about facets...

Adrien Grand (@jpountz) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17485355&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17485355)) To clarify, I was thinking of the case when a user sorts on a single field at search time, which happens to be the...

Adrien Grand (@jpountz) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17485378&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17485378)) I'm also wondering if this could help better support pre-filtering with NN vectors. We could use these sparse indexes to produce Bits representations of...

Adrien Grand (@jpountz) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17551429&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17551429)) Another potential use-case we are interested in for Elasticsearch would be to have the ability to visit a single document per field value for...

Ignacio Vera (@iverase) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17558103&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17558103)) I have been thinking on the ability if visiting a single documents per field value as explained for Adrien above and I think we...

Ignacio Vera (@iverase) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17558474&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17558474)) I open a draft PR that shows the idea I exposed above.

Robert Muir (@rmuir) ([migrated from JIRA](https://issues.apache.org/jira/browse/LUCENE-10396?focusedCommentId=17559724&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17559724)) If you just need the first document with the each value, why not use postings/TermsEnum?