[
https://issues.apache.org/jira/browse/LUCENE-5157?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13729643#comment-13729643
]
Robert Muir commented on LUCENE-5157:
-------------------------------------
I can say for sure I'm not happy with the API :) And its bogus its inconsistent
about subIndex/segment, so we should fix it one way or the other.
A few things i hate about the API for the record:
* should it really be in multitermsenum? if you have N-ord'enabled termsenums
(regardless if they come from DV or not), you could use this thing. on the
other hand the fact it uses termsenum at all could be seen as an impl detail of
merging (sorted)set dictionaries.
* i hate the 'object' owner and hairy slow-wrapper caching. on the other hand,
I felt it would be a trap to be anywhere else, because it can have real cost
and is not per-segment, so slow-wrapper feels like the right place.
So my comment about subindex had more to do with the first problem: do we think
it should be seen as a termsenum-merger or is it just a dv thing?
> Refactoring MultiDocValues.OrdinalMap to clarify API and internal structure.
> ----------------------------------------------------------------------------
>
> Key: LUCENE-5157
> URL: https://issues.apache.org/jira/browse/LUCENE-5157
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Boaz Leskes
> Priority: Minor
> Attachments: LUCENE-5157.patch
>
>
> I refactored MultiDocValues.OrdinalMap, removing one unused parameter and
> renaming some methods to more clearly communicate what they do. Also I
> renamed subIndex references to segmentIndex.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
