[
https://issues.apache.org/jira/browse/HBASE-17907?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16039656#comment-16039656
]
Enis Soztutar commented on HBASE-17907:
---------------------------------------
Thanks Ted for the review.
bq. For ScanResumerImpl::Resume(), what's the purpose for extra pair of curly's
?
That is an extra scope so that at the end of the scope the unique_lock for the
mutex is unlocked.
bq. The following local variable seems not used in ScanResumerImpl::Resume():
num_complete_rows tracking is from the Java-code. I've kept those although they
are not used for now so that when scan metrics are implemented, most of the
functionality will be there already.
bq. The above methods produce strings which seem to be identical for
corresponding states in controller and consumer. Consider distinguishing the
states with "controller" / "consumer".
It is function overloading with enum argument types. I don't think it is less
readable than the alternative.
bq. For the remaining elements in queue_, we can discard without further action
?
I think it is fine. When the result scanner is closed, you should not be able
to read from it.
I've addressed the other reviews, and added scanner_id to the other logs as
well.
> [C++] End to end Scans from Client/Table
> ----------------------------------------
>
> Key: HBASE-17907
> URL: https://issues.apache.org/jira/browse/HBASE-17907
> Project: HBase
> Issue Type: Sub-task
> Reporter: Enis Soztutar
> Assignee: Enis Soztutar
> Fix For: HBASE-14850
>
> Attachments: hbase-17907-v1.patch, hbase-17907_v2.patch,
> hbase-17907_v3.patch, hbase-scan-v0.patch
>
>
> Scan support from client, including the implementation for retrying RPCs,
> scanner management, scan result caching, prefetch, Client side interfaces and
> all related code.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
