lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hoss Man (JIRA)" <>
Subject [jira] Updated: (SOLR-1782) stats.facet assumes FieldCache.StringIndex - fails horribly on multivalued fields
Date Tue, 11 May 2010 01:03:39 GMT


Hoss Man updated SOLR-1782:

    Attachment: SOLR-1782.test.patch

patch containing two new test methods that demonstrate this bug -- one creates an index with
a single document (and two values) to trigger the AIOOBE, the second adds many docs with a
small number of distinct values, and demonstrates the incorrect stats facet values.

Off the top of my head i don't know a simple way to fix this ... if no one else has any better
suggestions, we should consider adding Schema based error checking to fail fast if the user
attempts to facet on fields which are not single valued string fields -- some other field
types *might* work, but i'd rather fail more often then we have to then return bogus statistics.

> stats.facet assumes FieldCache.StringIndex - fails horribly on multivalued fields
> ---------------------------------------------------------------------------------
>                 Key: SOLR-1782
>                 URL:
>             Project: Solr
>          Issue Type: Bug
>          Components: search
>    Affects Versions: 1.4
>         Environment: reproduced on Win2k3 using 1.5.0-dev solr ($Id: CHANGES.txt 906924
2010-02-05 12:43:11Z noble $)
>            Reporter: Gerald DeConto
>         Attachments: index.rar, SOLR-1782.test.patch
> the StatsComponent assumes any field specified in the stats.facet param can be faceted
using FieldCache.DEFAULT.getStringIndex.  This can cause problems with a variety of field
types, but in the case of multivalued fields it can either cause erroneous false stats when
the number of distinct values is small, or it can cause ArrayIndexOutOfBoundsException when
the number of distinct values is greater then the number of documents.

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:
For additional commands, e-mail:

View raw message