Adar Dembo has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/8127 )

Change subject: WIP: Pull StatusOr<T> and related tests from protobuf
......................................................................


Patch Set 1:

At first I thought this was supposed to be a StatusAnd<T>, which addresses a 
real pain point where some call paths want to return supplemental error 
information and it's nicer if that information was coupled with the Status 
itself. I remember Alexey distinctly wishing for that while working on a thorny 
CatalogManager error-handling patch.

But, this is just syntactic sugar to combine an OUT parameter into the return 
value, right? Is it even worth it?

I'm asking because it seems like with StatusOr<T> instead of one calling 
convention (e.g. "Status DoFoo(Bar* out1, Baz* out2)") we'll have two:

  Status DoFoo(const Bar& in) // functions without OUT params
  StatusOr<Bar> DoFoo(Baz* out2) // functions with 1+ OUT param

That is, functions with no OUT params will continue to use Status, while calls 
with at least one OUT param will return the first in the StatusOr and the 
remainder as normal OUT params.

I think this is especially weird for functions with multiple OUT params, 
because the first will be in StatusOr<T> while the rest will be regular OUT 
params. Or is the idea to combine all of them into StatusOr<pair>, 
Status<tuple>, or Status<some_struct>?

These concerns can be alleviated if I understood how we intend to apply 
StatusOr.


--
To view, visit http://gerrit.cloudera.org:8080/8127
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: If2d86a3769da733456956fb892de6d679a011a63
Gerrit-Change-Number: 8127
Gerrit-PatchSet: 1
Gerrit-Owner: David Ribeiro Alves <davidral...@gmail.com>
Gerrit-Reviewer: Adar Dembo <a...@cloudera.com>
Gerrit-Reviewer: Dan Burkert <danburk...@apache.org>
Gerrit-Reviewer: David Ribeiro Alves <davidral...@gmail.com>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Reviewer: Tidy Bot
Gerrit-Reviewer: Todd Lipcon <t...@apache.org>
Gerrit-Comment-Date: Thu, 28 Sep 2017 23:07:57 +0000
Gerrit-HasComments: No

Reply via email to