Tim Armstrong has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/16597 )

Change subject: IMPALA-10127: Fix performance for enforcement of 
lirs_tombstone_multiple
......................................................................


Patch Set 1: Code-Review+2

(4 comments)

http://gerrit.cloudera.org:8080/#/c/16597/1//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/16597/1//COMMIT_MSG@25
PS1, Line 25: a pointer to the front of the unprotected list (i.e. the
I found it a little counterintuitive that the front is the right end of the 
list. It was clearer when I read the code comments though.


http://gerrit.cloudera.org:8080/#/c/16597/1//COMMIT_MSG@38
PS1, Line 38: pathalogical
pathological


http://gerrit.cloudera.org:8080/#/c/16597/1//COMMIT_MSG@40
PS1, Line 40: 2000x
nice


http://gerrit.cloudera.org:8080/#/c/16597/1/be/src/util/cache/lirs-cache.cc
File be/src/util/cache/lirs-cache.cc:

http://gerrit.cloudera.org:8080/#/c/16597/1/be/src/util/cache/lirs-cache.cc@586
PS1, Line 586:   auto it = unprotected_tombstone_list_.begin();
I found this a bit counter-intuitive because I would've expected begin() to be 
the left-most element in the picture about, but it's the right-most. Maybe this 
is a normal way to illustrate queues but I found it a bit counter-intuitive 
from the point-of-view of C++ iterators.

Not sure that anything in the code or comments needs to change. Maybe 
explicitly show begin()/end() in the ASCII diagram if it doesn't make it too 
noisy?



--
To view, visit http://gerrit.cloudera.org:8080/16597
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I25b697f57c7daacccf8791a5a7b31878a6f7f1d2
Gerrit-Change-Number: 16597
Gerrit-PatchSet: 1
Gerrit-Owner: Joe McDonnell <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-Comment-Date: Wed, 14 Oct 2020 17:53:30 +0000
Gerrit-HasComments: Yes

Reply via email to