Todd Lipcon has posted comments on this change.

Change subject: KUDU-1259: new scanner API with an encapsulated Batch object
......................................................................


Patch Set 8:

(22 comments)

lots of the comments were on moved chunks of code rather than new code, so I 
didn't change those (to preserve history). If you feel like the cleanup's worth 
doing we can do another patch on top for those.

http://gerrit.cloudera.org:8080/#/c/1562/8//COMMIT_MSG
Commit Message:

Line 22: quite yet
> remove
Done


Line 22: soversion
> missing space
"soversion" is actually a term 
(https://cmake.org/cmake/help/v3.0/prop_tgt/SOVERSION.html)


Line 37: though
> don't need the "though" right?
Done


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/client/client.h
File src/kudu/client/client.h:

Line 980: A single KuduScanBatch instance may be reused for better performance
> not sure about this. how much of a performance diff is this really going to
It's better because it ends up re-using the same buffers for the responses, if 
you keep reusing the same scanbatch object. So, less allocator pressure and 
better cache locality.

The RPC controller is entirely embedded here - the user shouldn't have to worry 
about it. From the user's perspective it's pretty easy - just like any other 
"out-param" you can reuse it, and the reuse will replace the existing one, but 
reuse some of its storage (like how reusing a string will avoid having to 
realloc)


Line 981: KuduRowResult
> didn't you change this to something else?
Done


Line 982: derived from the batch
> how about: "previously obtained from batch"
Done


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/client/scan_batch.cc
File src/kudu/client/scan_batch.cc:

Line 33: 
> extra space
Done


Line 40: }
> can be one line (right? i forget the minutia of the style guide here)
Done


Line 43: "~scanbatch"
> remove (or add meaningful info)
woops, accidentally left in some debug


Line 70: "cell"
> clarify what you mean by "cell" maybe point to another one
this is just moved code, rather than new code


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/client/scan_batch.h
File src/kudu/client/scan_batch.h:

Line 64: NumRows
> docs
Done


Line 95:  
> weird wrapping
this and the following several comments are just moved code - would rather not 
edit it while moving it (so that git has a change of seeing the moved chunk)


Line 156:  
> docs since this is a public header
Done. I didn't doc the individual methods because they're obvious 
implementations of the iterator interface.


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/client/scanner-internal.cc
File src/kudu/client/scanner-internal.cc:

Line 424: }
> same q as before... can we avoid the extra line?
Done


Line 426: KuduScanBatch::Data::~Data() {
> same
Done


Line 442: addresses into absolute one
> this is a bit cryptic. How about: First, rewrite the relative offsets into 
moved code


Line 465: 
> extra lines
Done


Line 470: KuduRowResult
> why still use KuduRowResult here?
Done


Line 476: Extracted 0 rows
> more info or remove (or increase VLOGness)
moved code


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/client/scanner-internal.h
File src/kudu/client/scanner-internal.h:

Line 171: 
> extra lines
Done


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/rpc/rpc_controller.cc
File src/kudu/rpc/rpc_controller.cc:

Line 35: Canot
> typo
Done


http://gerrit.cloudera.org:8080/#/c/1562/8/src/kudu/rpc/rpc_controller.h
File src/kudu/rpc/rpc_controller.h:

Line 49: Swap
> docs
Done


-- 
To view, visit http://gerrit.cloudera.org:8080/1562
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I29fd4fbb8b906ffa591853ab625ac4b089da4bc9
Gerrit-PatchSet: 8
Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-Owner: Todd Lipcon <[email protected]>
Gerrit-Reviewer: Binglin Chang <[email protected]>
Gerrit-Reviewer: Dan Burkert <[email protected]>
Gerrit-Reviewer: David Ribeiro Alves <[email protected]>
Gerrit-Reviewer: Internal Jenkins
Gerrit-Reviewer: Martin Grund <[email protected]>
Gerrit-Reviewer: Todd Lipcon <[email protected]>
Gerrit-HasComments: Yes

Reply via email to