[
https://issues.apache.org/jira/browse/HBASE-19057?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16213979#comment-16213979
]
Zheng Hu commented on HBASE-19057:
----------------------------------
[~anoop.hbase], Thanks for your reply. Agree your first 4 points.
point #2 is my mistake. Thanks for the careful review.
point #4 is a nice optimization , *_NEXT_COL and *_NEXT_ROW are the same
case, I think.
Some discussion about the last 2 points:
point #5
> I dont think in any of the FL filterRowKey, we should call filterAllRemaining.
Sounds good, the filterRowKey of FilterList is false , it indicate that they
exist at least one sub-filter whose filterRowKey is false, However for those
sub-filters whose filterAllRemaining are true, their filterRowKey are
definitely true. So filterAllRemaining can be removed.
> Also can early out than just set retVal and continue, as in AND filter
Actually, we have some type of filter , such as PrefixFilter, they have some
state in their filterRowKey implementation, so if we return early, those
filters will not have a chance to update they state, which lead to the failure
result. UT indicate that.
point #6
> Got now why no seekHintFilter set here. As that has to be all filters.
Yes, only when all filters return seek_hint, our FL return seek_hint.
> But other things abt filterAllRemaining is still valid
Sounds good.
> Merge branch HBASE-18410 to master branch
> -----------------------------------------
>
> Key: HBASE-19057
> URL: https://issues.apache.org/jira/browse/HBASE-19057
> Project: HBase
> Issue Type: Sub-task
> Components: Filters
> Reporter: Zheng Hu
> Assignee: Zheng Hu
> Fix For: 2.0.0-beta-1
>
>
> Open this issue to fix conflict , run HadoopQA and gather other feedback.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
