[
https://issues.apache.org/jira/browse/LUCENE-5480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13916175#comment-13916175
]
Robert Muir commented on LUCENE-5480:
-------------------------------------
I think the current bug is a longstanding one, because prefix and suffix
stripping is not intertwined (so continuation classes from prefixes dont apply
to suffixes and so on).
This causes overstemming today.
I'd like to fix the current bug(s) here with the uploaded patch and open a
followup issue for that... its progress.
> Hunspell shouldnt merge dictionary entries
> ------------------------------------------
>
> Key: LUCENE-5480
> URL: https://issues.apache.org/jira/browse/LUCENE-5480
> Project: Lucene - Core
> Issue Type: Bug
> Components: modules/analysis
> Reporter: Robert Muir
> Attachments: LUCENE-5480.patch
>
>
> Ive been writing lots of little unit tests for this thing, and I'm pretty
> positive i screwed this up in LUCENE-5468... sorry
> Otherwise the whole "prefix-suffix dependencies" described in the manpage
> won't work.
> Either 'words' should be changed from FST<Long> to FST<IntsRef>, or when
> there are duplicates we should add 'padding' that we just consume
> (suggester-style). The latter is a little tricky, but I think this is
> generally uncommon so it would keep the FST smaller.
> shouldnt be hard to fix.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
