Re: [gentoo-dev] [last rites] virtual/x11

[Ciaran McCreesh]

On Mon, 26 Mar 2007 07:52:25 -0700 (PDT)
"Alec Warner" <antarus@g.o> wrote:
> > On Sun, 25 Mar 2007 23:07:03 -0400
> > Seemant Kulleen <seemant@g.o> wrote:
> >> Ciaran has brought attention to a very important thing -- QA seems
> >> to take a backseat to a few things, and it is actually a little
> >> disturbing that it does.
> >
> > I believe the QA team expect developers to *ask* when they want QA
> > to intervene on a thing like this. There aren't enough people in QA
> > to find every problem in the tree on their own...
>
> There is no motivation to ask if the developer in question doesn't
> care about breaking the tree in the first place.

It doesn't have to be the developer in question who brings the issue to
QA's attention...

The problem, as I understand it, is this:

The tree is big. There are lots of changes. Some of these changes are
breakages. Some of these breakages can be detected automatically, which
QA are fairly good at doing. Some of these breakages can't be, and QA
won't know about these until someone reports it to them.

Now, the tricky part. Some breakages, like this one, are the result of
things that aren't directly tree changes, but rather an ongoing
general change. Developers are generally expected to do the maintenance,
but after a certain point, it's no longer worth waiting for them. QA can
take over or authorise someone to take over in these situations, but
only if an interested party steps up and says to QA "it's about time
something gets done about this"...

There's this general feeling that QA are expected to know everything
and deal with every breakage. That isn't realistic, and I strongly
suspect it's being used by certain people as a way of passing blame
onto someone else. QA is, first and foremost, the responsibility of
package maintainers. Secondly, it's the responsibility of arch teams.
Only if both of those fail should the QA team get involved.

-- 
Ciaran McCreesh