On Sep 28, 2010, at 5:20 PM, Dmitriy Ryaboy wrote:
A couple of notes:
I've been operating under the assumption that at least 1 "+1"
committer vote
(not including author of contribution) is required for codebase
changes.
This is a sensible rule imho.. but apparently not a real one :).
Thoughts?
I think this is what is meant by saying a code change requires lazy
consensus. The rules we've really been running under is that any code
changes must have at least one +1 besides the author. If the author
is a committer, we have not required the other +1 to be one. But if
the author is not a committer we have required that the +1 come from a
committer. Making this clear would be good.
Vetoes in response to a commit message -- seems like one should be
able to
veto *before* a commit message, too. Nobody likes reverting codebases.
Yeah, we never check in code over a -1. Again, I can make this more
clear. Reverting code is like going to war. By the time you're there
you know something went way wrong to get you there.
Inactive / Emeritus committers -- does everyone listed as initial
committer
actually pass the 6-month requirement? I think Ben Reed and Giri
might not,
which would make things awkward right off the bat.
Before I proposed the initial PMC list I checked that everyone on the
proposed list passed the test (including Ben and Giri). That's why I
went through and had a bunch of people moved to the emeritus list
before I proposed moving Pig to a TLP.
Distinction between PMC and Committer list -- right now they are
identical,
correct? Are they normally very different in other projects? Is that
the
sort of thing that develops over time, and the initial groups tend
to be the
same?
Different projects handle it differently. I would propose that we
maintain a distinction. If we don't I feel it will be hard to add
committers, because at least for me the bar for PMC is higher than the
bar for committers.
Alan.
Thanks,
-Dmitriy
