Re: Priority-Regression policy

[Patrick McCarty]

On 2009-11-30, Graham Percival wrote:
> On Sun, Nov 29, 2009 at 07:34:37PM -0800, Patrick McCarty wrote:
> > On 2009-11-29, Graham Percival wrote:
> > > What's the feeling amongst developers about what should be ranked
> > > as priority-Regression (and thus stop a release) ?  In particular,
> > > should *everything* that used to work -- even if it was by
> > > accident? -- be ranked a Regression?
> > 
> > Maybe we could add labels indicating which release an issue blocks?
> 
> The previous policy, which I assume stands, is that anything
> ranked Priority-Regression is a "release blocker".  IIRC, at one
> point this even blocked unstable releases.

Ah, okay.  Just as long as we decide not to block unstable releases,
I'm fine with it.  That's asking a little too much.  :-)

> > > I don't particularly mind which way we decide, but I'd like it to
> > > be consistent, and I'm going to insist that if something is
> > > Priority-Regression, it blocks a release.
> > 
> > IMO, regressions from 2.13 should get first priority and should block
> > 2.14, but other regressions should be considered on a case-by-case
> > basis.
> 
> I'm not opposed to this, although if we want to go this route, I
> propose *removing* the Priority-Regression label.  We could then
> use High, Medium, Low, Postponed.  Regressions would then be
> High-priority by default, but developers could lower it if the
> regression was due to an architecture change, or if it only worked
> by accident originally.

What if we use "Type: Regression" instead of "Priority: Regression",
and then label each with a priority of either High, Medium, Low, or
Postponed (assuming we decide to go this route)?

Thanks,
Patrick