[
https://issues.apache.org/jira/browse/MAHOUT-706?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13044329#comment-13044329
]
Sean Owen commented on MAHOUT-706:
----------------------------------
I can commit this; I imagine you certainly know what you're doing but are there
any thread-safety concerns here? looks like the reused object is
thread-local... haven't thought through the implications though.
> reuse lucene tokenstreams
> -------------------------
>
> Key: MAHOUT-706
> URL: https://issues.apache.org/jira/browse/MAHOUT-706
> Project: Mahout
> Issue Type: Improvement
> Reporter: Robert Muir
> Assignee: Grant Ingersoll
> Priority: Minor
> Fix For: 0.6
>
> Attachments: MAHOUT-706.patch
>
>
> Currently, mahout uses Lucene's non-reusable analysis API.
> This means that per-"document", a lot of objects are recreated (e.g. every
> TokenStream in the analysis chain, every Attribute).
> This can create a lot of unnecessary overhead, particularly if "documents"
> are short.
> It looks like an easy win to use the reusable API (reusableTokenStream)
> instead.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
