[jira] [Commented] (SOLR-11372) enable refinement testing in TestCloudJSONFacetJoinDomain

["Hoss Man (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SOLR-11372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16172349#comment-16172349 ] 

Hoss Man commented on SOLR-11372:
---------------------------------

this should be fairly straight forward, but I would suggest doing it in stages.....

1) replace the current {{FACET_LIMIT = 1 + ...}} constant with a randomized value that is sometimes more/less then the #unique values, and add the {{refine}} param to every generated facet.

2) Assuming that works under beasting for a but and no obvious problems jump out, we can work dowards increase the randomization & edge cases caught by starting with some refactoring...

* add a variable the facet modeling class to support a unique limit for each facet - default to using the global FACET_LIMIT
** update the json serialization to add refine:true if and only if the limit specified is < #unique values

3) once that refactoring is vetted, then...

* the code that generates the randomized facet options can be updated to randomize the limit as well
* the refine param can also be randomized a bit more:
** if the limit is > #unique values we can randomly pick any refine value (true, false, or unspecified) and it should still produce valid results
** if the random limit is < #unique values, we can also randomly pick a refine value -- and if we pick unspecified or false, setting the overrequest param > #unique values should also still make the test pass
** if (for any reason) we randomize refine=true we should be able to pick a random overrequest param < #unique values w/o causing test failures.

> enable refinement testing in TestCloudJSONFacetJoinDomain
> ---------------------------------------------------------
>
>                 Key: SOLR-11372
>                 URL: https://issues.apache.org/jira/browse/SOLR-11372
>             Project: Solr
>          Issue Type: Test
>      Security Level: Public(Default Security Level. Issues are Public) 
>            Reporter: Yonik Seeley
>
> This test has great random tests that ensure that the count returned for a bucket matches the number of documents returned from an equivalent filter.  We should enable randomly testing smaller limits in conjunction with refinement to ensure we still get accurate counts for buckets that are returned.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

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