Re: [gentoo-dev] QA Roles v2

[Paul de Vrieze]

On Thursday 02 March 2006 14:09, Mark Loeser wrote:
> Paul de Vrieze <pauldv@g.o> said:
> > * Just because breaking policy breaks a QA tool, but is guaranteed to
> >   never break itself (formatting policy, like space vs. tab etc.)
> > does not increase the severity of the breakage.
>
> I had hoped something like this would have just been understood to not
> be too severe, since it doesn't really break anything but coding
> standards.
>
> > * Before any enforcement is possible, QA must establish a well
> > supported (debated on dev-) exception policy. While it were nice if
> > exceptions are not needed, reality is that they can not be avoided.
> > Therefore there must be an exception policy.
>
> I'm not sure what you mean here.  You mean for each instance?  In
> general?  In general can be difficult since it leaves a lot of things
> up for interpretation.  For each instance, 99% of the time an
> acceptable interim solution should be able to be achieved between the
> QA team and the maintainer.  In situations where we can't figure out
> how to best address the situation, opening the discussion up to -dev
> may help, but in the end it should come down to an agreement between
> the maintainer and the team.

The policy should be general. It could be something like "Developer and QA 
discuss the exception and the solution to be used. If they do not agree 
the council.... . In any case when an exception is made, appropriate bugs 
are created, including a bug to request a feature that makes the 
exception unneeded. When the requested feature is available and stable, 
the exception becomes invalid and the feature must be used instead."

My idea is that QA can not enforce controversial things before such a 
policy exists. Otherwise exceptions stay only a theoretical possibility, 
and arguments continue.

Paul

-- 
Paul de Vrieze
Gentoo Developer
Mail: pauldv@g.o
Homepage: http://www.devrieze.net