PATCH: Run V3 tests on installed compiler

[Mark Mitchell]

Benjamin Kosnik wrote:
> The
> following is just feedback to Mark on his maintainer style. Hopefully it
> is useful.

Thank you.

>>| I can
>>| certainly remember to wait 24 hours in future, but I'm not sure how
>>| everyone else will know.
>>
>>well, that is not the first time people would send ping messages to
>>the list, if it ever happens that your patch is unnoticed...
> 
> 
> It seems unlikely that this patch would have not been reviewed. However,
> if it was, with GWP and clearly given notice, I would not have had a
> problem with this patch being checked in after the time deadline
> expired. 

I didn't actually mean "how will people know the patch is there?"; I 
meant, how will maintainers who are not knowledgeable out the ways of V3 
know what's acceptable?

> We have peer review for a reason.... GWP means you have the ability to
> check in anywhere. Not that you should check in without consulting area
> maintainers. That's what I mean when I say professional courtesy, but
> perhaps my view of things is actually not reflective of reality.

How I've heard it put is "it's OK to check in where you feel confident 
that you know what you're doing."  I did feel confident about this patch.

(I think that the root cause of the problem here is that I viewed this 
change as a minor bit of clearly non-controversial maintenance; the V3 
people clearly viewed as a major paradigm shift.  Probably if it had 
been as minor a change as I believed, there would have been no major 
problem -- although I do understand, and accept, your request that all 
V3 patches be posted for 24 hours before check-in.)

FWIW, the level of professional courtesy I expect as a C++ maintainer is 
that if I object to a patch that has gone in, then a discussion will be 
had, and, ultimately, the patch might be reverted.  That's why I 
immediately offered to revert my patch, once you objected.  But, I don't 
feel it discourteous for GWP people to check in changes without 
consulting the C++ maintainership first, so long as they go through the 
usual testing procedures.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304