[ 
https://issues.apache.org/jira/browse/SOLR-15261?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17302150#comment-17302150
 ] 

Michael Gibney commented on SOLR-15261:
---------------------------------------

This failure seems consistent. FWIW I don't see {{SortField.Type.STRING_VAL}} 
used anywhere in Solr but in two tests:
 # org/apache/solr/uninverting/TestFieldCacheSortRandom.java
 # org/apache/solr/uninverting/TestFieldCacheSort.java

I'm not sure why only the former seems to yield failing tests as a result of 
this change; but in any case both of these seem to be "uninverted" variants of 
Lucene-side tests that were [largely removed as part of 
LUCENE-9796|https://github.com/apache/lucene/commit/f3a284ad830b9473467b5cb364408db5e4cc607f#diff-ddd870424add53ef5b85f7243cdabb805f57e472468a3eb14b54f2f51aa56844].
 If Solr takes its cues from what happened to the analogous Lucene-side tests 
(and in this case I think that seems appropriate, unless there are some hidden 
references to STRING_VAL that I'm not seeing?), then maybe these tests 
targeting "STRING_VAL" can be switched to apply only when the underlying dv 
type is _actually_ BINARY (assuming this is a supported use case), as opposed 
to "SORTED masquerading as BINARY"? I think that may be essentially what Robert 
Muir is suggesting in [this 
comment|https://issues.apache.org/jira/browse/LUCENE-9796?focusedCommentId=17301820#comment-17301820].

> SortedDocValues no longer extends BinaryDocValues
> -------------------------------------------------
>
>                 Key: SOLR-15261
>                 URL: https://issues.apache.org/jira/browse/SOLR-15261
>             Project: Solr
>          Issue Type: Task
>            Reporter: Christine Poerschke
>            Assignee: Bruno Roustant
>            Priority: Major
>             Fix For: main (9.0)
>
>          Time Spent: 1h 20m
>  Remaining Estimate: 0h
>
> From a quick look it appears some changes are needed in 
> https://github.com/apache/solr to adjust to the LUCENE-9796 changes. Couldn't 
> find an existing ticket for it and so creating this one (though I don't 
> personally have any specific plans to work on this).



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to