[ 
https://issues.apache.org/jira/browse/LUCENE-5229?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13773108#comment-13773108
 ] 

Robert Muir commented on LUCENE-5229:
-------------------------------------

{quote}
I wasn't confused by the fact that I received NaN, only pointed out that when 
you use Expression, the result is not in the 'score' field, but the 'field' 
field.
{quote}

You invoked IndexSearcher.search(query, filter, n, *Sort*) and you were 
surprised that the result of the sort goes there?

I think this kinda stuff only furthers to reinforce my argument that this stuff 
is way too specialized and complicated.

                
> remove Collector specializations
> --------------------------------
>
>                 Key: LUCENE-5229
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5229
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Robert Muir
>            Assignee: Robert Muir
>
> There are too many collector specializations (i think 16 or 18?) and too many 
> crazy defaults like returning NaN scores to the user by default in 
> indexsearcher.
> this confuses hotspot (I will ignore any benchmarks posted here where only 
> one type of sort is running thru the JVM, thats unrealistic), and confuses 
> users with stuff like NaN scores coming back by default.
> I have two concerete suggestions:
> * nuke doMaxScores. its implicit from doScores. This is just over the top. 
> This should also halve the collectors.
> * change doScores to true by default in indexsearcher. since shai was 
> confused by the NaNs by default, and he added this stuff to lucene, that says 
> *everything* about how wrong this default is. Someone who *does* understand 
> what it does can simply pass false.

--
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: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to