Todd Lipcon has posted comments on this change.

Change subject: [c++ client] - Expose the schema being scanned in KuduScanBatch
......................................................................


Patch Set 4:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/2020/4/src/kudu/client/scan_batch.h
File src/kudu/client/scan_batch.h:

Line 171:   const KuduSchema* client_schema_;
hrm, I have some worries about the perf impacts of this. keeping the RowPtrs is 
small makes them cheap to pass/copy by value, but the wider they get, the less 
they're like a simple pointer. What's the use case where you need this pointer 
from the RowPtr level and not from the ScanBatch level?

The comments during the review seemed to indicate there isn't really an 
argument here except for cross-client consistency, which I dont think is that 
strong of a reason given it adds an unavoidable per-row perf overhead to the 
scan path


-- 
To view, visit http://gerrit.cloudera.org:8080/2020
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: Id4276036d5ba7faf7959b4324fd013b68a08bec5
Gerrit-PatchSet: 4
Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-Owner: David Ribeiro Alves <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: David Ribeiro Alves <[email protected]>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Reviewer: Todd Lipcon <[email protected]>
Gerrit-HasComments: Yes

Reply via email to