[
https://issues.apache.org/jira/browse/LUCENE-6105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14241295#comment-14241295
]
Robert Muir commented on LUCENE-6105:
-------------------------------------
I don't disagree with the bug aspect, and i am sure the patch is great.
We use these FSTs in lots of code so the possibility of a performance bug
(separate from correctness issue) exists. We should at least wait for a few
iterations of jenkins, ideally some nightly benchmark reports, given how
widespread FST usage is, before getting it out to users. This stuff happens
automatically with time, but just needs some time.
> Don't create root arc cache for tiny FSTs
> -----------------------------------------
>
> Key: LUCENE-6105
> URL: https://issues.apache.org/jira/browse/LUCENE-6105
> Project: Lucene - Core
> Issue Type: Bug
> Components: core/FSTs
> Reporter: Michael McCandless
> Assignee: Michael McCandless
> Fix For: Trunk, 5.x
>
> Attachments: LUCENE-6105.patch
>
>
> The purpose of the root arc cache is to speed up lookups for ASCII terms, but
> it adds high overhead if the FST is already tiny.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
