[
https://issues.apache.org/jira/browse/LUCENE-6000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14164810#comment-14164810
]
Ryan Ernst commented on LUCENE-6000:
------------------------------------
I checked in as is (with the deletion of course), but I don't have a problem
with continuing this work to nuke the Impl classes. The only caveat I see is
this would require embedding a small amount of java code (100+ lines or so?)
into the jflex files. This may not be ideal if/when the tokenizer api is
tweaked, and this functions need to be iterated on. But that is so rare it
probably doesn't matter. Or maybe I am missing something and it is simpler
than it looks to me. Either way, if you feel getting rid of the separate impls
will make this even cleaner, then we can open another issue and I'm happy to
help there.
> Remove StandardTokenizerInterface
> ---------------------------------
>
> Key: LUCENE-6000
> URL: https://issues.apache.org/jira/browse/LUCENE-6000
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Ryan Ernst
> Fix For: 5.0, Trunk
>
> Attachments: LUCENE-6000.patch
>
>
> This interface existed for backcompat, so that each impl had at least some
> common minimal interface, and could be used by StandardTokenizer. However,
> in LUCENE-5999 backcompat for standard tokenizer was implemented using
> separate named classes. We should remove this interface, as it no longer
> serves a purpose.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
