On Thu, May 21, 2015 at 7:05 PM, Stas Malyshev <smalyshev(a)wikimedia.org> wrote:
Neither should
you, that is the point of code review. Then again, if there
In theory, this is true. In practice, there are numerous occasions where
people have to self-+2 - typoes, forgotten files, CI glitches, rebases,
etc. Well, ok, "have to" is a strong word here - all of it can be worked
around by dragging in somebody and asking them "please +2 this" - but
again, that would be working against the setup and also training people
that the system sucks and they have to work around it to be effective.
For better or worse, self+2 is not uncommon in Ops workflow today in
repos like ops/puppet. We really do try to get other-reviews when
we're unsure about things (or they're complicated, or involve new
design work, etc...) but there are also many common cases where simple
things need doing and nobody can be bothered interrupting themselves
and someone else to sync up on review. It's a question of trusting
+2-ers with sound judgement: they should know when self+2 is a
reasonable option and when it isn't (but good luck documenting every
possible rationale for it). Usually the induced level of shame for a
self+2 that breaks things is sufficient feedback on judgement.