Bug#625449: Permanent BSP patch
On 03/05/11 at 22:27 +0900, Charles Plessy wrote:
> [DEP1 drivers CCed].
>
> Le Tue, May 03, 2011 at 01:25:49PM +0100, Neil McGovern a écrit :
> > Package: developers-reference
> > Tags: patch
> >
> > [0] http://lists.debian.org/debian-devel-announce/2011/03/msg00016.html
>
> > Index: pkgs.dbk
> > ===================================================================
> > --- pkgs.dbk (revision 8790)
> > +++ pkgs.dbk (working copy)
> > @@ -1947,6 +1947,11 @@
> > <itemizedlist>
> > <listitem>
> > <para>
> > +Upload fixing only release-critical bugs older than 7 days, without maintainer activity for 7 days: 0 days
> > +</para>
> > +</listitem>
> > +<listitem>
> > +<para>
>
> Hi all,
>
> in light of the GCC mass bug filing, I just would like to say my unease with
> the pressure given by the tight delays that maybe will be advertised in the
> Developers Reference.
>
> Fixing these bugs is more than just applying a patch, it is also forwarding the
> issue and discussing with Upstream. In my case, I also make the voluntary
> choice to look in Launchpad and fix their GCC bug through my changelogs. And
> our VCS, writable by all DDs, often contain changes in preparation that would
> be ignored by a NMU.
>
> The necessity to care was brilliantly reminded by Neil in the announcement
> cited above, but with that patch applied the Developers reference would read:
>
> Unless you have an excellent reason not to do so, you must then give some time
> to the maintainer to react (for example, by uploading to the DELAYED queue).
> Here are some recommended values to use for delays:
>
> - Upload fixing only release-critical bugs older than 7 days, without
> maintainer activity for 7 days: 0 days
>
> I feel it recommending a zero-day delay does not carry the same message as
> “nobody prevents you from uploading NMUs to DELAYED”, from the link above.
I agree that the resulting wording of patch is suboptimal, and that
recommending 0-day NMUs is not the way to go. We are rarely in need for
action in less than a couple of days in Debian, so the current policy
seems fine to me.
Also, I question the authority of the release team for deciding NMU
policies on their own.
> One way to reduce the pressure would be to ask the NMUer to take into account
> the state of the VCS where the work is being done (debcheckout -a), and to
> commit his changes after upload.
>
> Would an update to the NMU section of the Developers Reference to better
> mention VCSes be welcome, or is it considered part of DEP1 and therefore need
> to be discussed more widely ?
With that, I disagree. There are many different ways to interact
with the VCS. For example, the VCS could contain additional changes.
Should they be uploaded together with the NMU? In some cases, it might
make sense, but what if the committed-but-not-uploaded changes are crap?
So before interacting with the package's VCS, I think that it's better
to require, or at least strongly encourage, discussion with the
maintainer(s).
- Lucas
Reply to: