Re: Combine four insns

[Mark Mitchell <mark at codesourcery dot com>]

Diego Novillo wrote:

>> I don't want to see it get too procedural to get something checked in.
>> I'd rather have a culture where it's not too hard to get things in, but
>> where we are responsive to problems raised after the patch is checked
>> in.
> 
> Agreed.  I'm not very concerned about getting patches in that may break
> something, even if it may not be immediately obvious at the time.  I
> would much rather roll patches back instead of agonizing that every last
> bit of a given patch meets some impossible standard of perfection.

Right.  I think that in the past we've tried to use patch approval as a
way of dealing with the problem of people who check something in and
then disappear.  We try to make the patch perfect up front so that we
don't have to deal with a patch that's been abandoned.

I'm not saying that's not a problem; sometimes people *have* abandoned
patches and that's bad.  But, we know that in many cases we can trust
people to clean up after themselves.  Many contributors have a strong
track record of doing that, and for contributors affiliated with a
corporation it's reasonable to ask others at the corporation to fix
problems, even if the original contributor is MIA.

So, I think we should use reasonable judgment.  I'm all for review, but
I think it's more important that people be willing to deal with the
inevitable post-commit problem than that we strictly follow any
particular pre-commit procedure.

-- 
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713