Re: Quality Assurance Group mini-policy
[was it really neccessary to crosspost to three lists?]
On Mon, Jun 07, 1999 at 06:36:58AM -0700, Craig Brozefsky wrote:
> > However, if QA Group members make 3 consecutive bugfix uploads within two
> > months, with still no action from the actual package maintainer, then the
> > package will be marked orphaned, and the 'Maintainer' field of the
> > package will be set to "Debian QA Group <email@example.com>".
> I am not sure we want to impose such a practice on developers. The
> term "forced orphaning" is not intended to give violent connotations
> to the act of marking packages as orphaned.
> 1. "forced orphaning" by the QA group can occur in as little as 3
> days, a long weekend. Assume we have 3 critical bugs, reported a
> day apart from one another. Since the required response time is 2
> days for critical bugs before a QA team upload, you could have as
> many as three uploads in a 6 day period, resulting in the package
> being marked as orphan. If we are in a freeze, this is halved, so
> three days, three uploads.
> 2. NMU policies are sufficient to guarantee quality, we do not need to
> extend this into something covering "forced orphaning". QA
> mini-policy should have no bearing on Debian's expectations of
> developers as maintainers, as there is no need.
> >From what I can tell, current policy encourages waiting several weeks
> before doing even an NMU to unstable (regardless of bug severity).
> This mini-policy would be a fairly major change. Even if it's not
> official policy or holds no real wait, it does indeed present a
> different understanding of what maintainers need to do to retain their
> maintainership of a package than the rest of the policy documents, and
> should be re-examined.
First of all, I think one thing is not being clear (probably because
the text is not vaguely written, but hey...). The expression "if this
and that, then we'll do this and that" does not mean that we will be
obligated to proceed with the specified action immediately. It just
gives the power to the QA Group to do so.
Nobody thinks that after <whatever> days the QA group member who
submitted a wishlist bug report will do the upload with the bug fixed.
The text's intent is just to give that opportunity to the QA group
members (in this specific case, a consultation with the rest of the
group would be most welcome. I'll add that to the text).
Currently, they have no right to do absolutely anything on their own,
and we want to change that.
The forced orphaning (as you call it) is something we need, since there
can be packages that can currently rot for a couple of releases (yes!)
in the archive, and nobody has the right to touch them, since nobody
knows where is the maintainer and will he mind. Take hwtools as an
example. C. Lameter orphaned the package a long time ago, after some
time S. Brentrup took it, an then dissappeared from the face of the earth.
And what can we do? Nothing. Only after J. Gunthorpe made a check on the
PGP keys and accounts on master, that we determined that SB is gone,
and could mark the package orphaned. If that hadn't happened, the
package would stay intact for another release... and God knows how
Did I mention that I will accept any patches you (or anyone else) makes
against the text? Output of `diff -u` is preffered.
enJoy -*/\*- http://jagor.srce.hr/~jrodin/