[jira] [Commented] (LUCENE-8041) All Fields.terms(fld) impls should be O(N) not O(log(N))

["Robert Muir (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-8041?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16240977#comment-16240977 ] 

Robert Muir commented on LUCENE-8041:
-------------------------------------

It doesn't need to be all *all* fields.terms impls. It is enough to optimize the default codec. 

TreeMap is a good simple default, all the various alternative terms dicts can continue to use it.
But the default codec should optimize for the access behavior that matters: accessing a field randomly.

I don't think we should remove field iteration/Fields unless we remove the ability to change term vectors "per-doc". It is currently needed (e.g. by CheckIndex) to know what fields were truly indexed for a specific document with vectors, since that may disagree with FieldInfos. If we fixed that, then it would truly be unnecessary and FieldInfos would be all we need.


> All Fields.terms(fld) impls should be O(N) not O(log(N))
> --------------------------------------------------------
>
>                 Key: LUCENE-8041
>                 URL: https://issues.apache.org/jira/browse/LUCENE-8041
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: David Smiley
>
> I've seen apps that have a good number of fields -- hundreds.  The O(log(N)) of TreeMap definitely shows up in a profiler; sometimes 20% of search time, if I recall.  There are many Field implementations that are impacted... in part because Fields is the base class of FieldsProducer.  
> As an aside, I hope Fields to go away some day; FieldsProducer should be TermsProducer and not have an iterator of fields. If DocValuesProducer doesn't have this then why should the terms index part of our API have it?  If we did this then the issue here would be a simple transition to a HashMap.
> Or maybe we can switch to HashMap and relax the definition of Fields.iterator to not necessarily be sorted?
> Perhaps the fix can be a relatively simple conversion over to LinkedHashMap in many cases if we can assume when we initialize these internal maps that we consume them in sorted order to begin with.



--
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