[jira] [Updated] (SOLR-4310) If groups.ngroups is specified, the docList's numFound should be the number of groups

["John Wooden (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/SOLR-4310?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

John Wooden updated SOLR-4310:
------------------------------

    Attachment: SOLR-4310_4.patch

Added another test class to check a few more use cases, both single-core and distributed.  There are also test cases written that rely on SOLR-2894 which are commented-out.

Fixed an issue where numFound (group count) would be 0 for a single-core edge case.

Our tests suggest that distributed works just fine with 4310 unless Amit recalls what he alluded to previously.

On a semi-related note....

When writing the additional tests, we noticed some inconsistent behavior around rows. Not a result of the 4310 patch nor pertinent to 4310's purpose, just something we discovered.

On a single core with group.limit > 1 and group.main, setting rows=10 will return 10 _documents_. A distributed setup with the same params will return 10 _groups_. A commented-out failing testcase is included in the patch. If others can confirm it, we can open a new JIRA ticket for it.
                
> If groups.ngroups is specified, the docList's numFound should be the number of groups
> -------------------------------------------------------------------------------------
>
>                 Key: SOLR-4310
>                 URL: https://issues.apache.org/jira/browse/SOLR-4310
>             Project: Solr
>          Issue Type: Improvement
>          Components: search
>    Affects Versions: 4.1
>            Reporter: Amit Nithian
>            Assignee: Hoss Man
>            Priority: Minor
>             Fix For: 4.4
>
>         Attachments: SOLR-4310_2.patch, SOLR-4310_3.patch, SOLR-4310_4.patch, SOLR-4310.patch
>
>
> If you group by a field, the response may look like this:
> <lst name="grouped">
> <lst name="series">
> <int name="matches">138</int>
> <int name="ngroups">1</int>
> <result name="doclist" numFound="138" start="0">
> <doc>
> <int name="id">267038365</int>
> <str name="name">
> Larry's Grand Ole Garage Country Dance - Pure Country
> </str>
> </doc>
> </result>
> </lst>
> </lst>
> and if you specify group.main then the doclist becomes the result and you lose all context of the number of groups. If you want to keep your response format backwards compatible with clients (i.e. clients who don't know about the grouped format), setting group.main=true solves this BUT the numFound is the number of raw matches instead of the number of groups. This may have downstream consequences.
> I'd like to propose that if the user specifies ngroups=true then when creating the returning DocSlice, set the numFound to be the number of groups instead of the number of raw matches to keep the response consistent with what the user would expect.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org