Bug#850156: Please firmly deprecate vendor-specific series files [and 1 more messages]
Had this sitting here half drafted, but as I got poked privately also
due to the apparent incoherence in the first part, I'm sending the
reply for that now. And will handle the other part later on.
Although, I guess, because it might be only partly on topic, I'm
considering whether it might be better to do more extensive write-ups
elsewhere on my website or similar, to articulate it properly, and
then possibly link it here, as then I'd also have reference to point
to. Because I do realize (or at least that seems apparent), mine is
probably a very minority view within the project, and that people
going on and on on this kind of tirades and ramblings become annoying
and/or tiresome fast. So then I could try to stop interjecting on the
subject, while evaluating how much these situation bother me until
they become unbearable, at which point I'd just disengage and distance
myself from work involving them.
On Tue, 2018-07-31 at 19:43:32 -0700, Russ Allbery wrote:
> Guillem Jover <firstname.lastname@example.org> writes:
> > If someone wants to see dpkg changed in some way related to this, I'd
> > request the same thing I did to Ian a couple of years ago, gather input
> > from derivatives and other current users, on their reasons for using it,
> > or start a discussion with them on whether they'd be satisfied with
> > potential alternatives, etc.
> At least from a Policy perspective, I don't see any obvious need to remove
> the feature from dpkg regardless of the outcome of this discussion (nor do
> I think that's within the scope that Policy should even consider). dpkg
> may support all kinds of things that Debian chooses not to use or even
> prohibit in the archive, and I think that division is healthy and good.
> dpkg is a more general tool than just Debian.
> dgit is *also* a more general tool than just Debian, of course, and I can
> see some reasons why the dgit maintainers may aspire to handling every
> package that dpkg can handle, even ones that wouldn't be uploaded to
> Debian, and thus may care about whether the feature is supported at all.
> But I think that's outside the scope of Policy and could be discussed
> betweeen dpkg and dgit maintainers if that's a concern.
> > That Ubuntu finds it to be a problem as a Debian downstream, with these
> > packages percolating into Ubuntu? Well, you could also have tried to
> > argue your case to the ftp-masters and lintian maintainers, that this
> > is making your life difficult, and whether they would reject packages
> > with debian/patches/ubuntu.series. Or convince the maintainers in
> > Debian using them that this is in fact not helping you.
> > Apparently, you do not even need to do that anymore, you just need to
> > get a ctte to ban them from Debian.
> I'm a bit confused about why you're upset with us asking the Technical
> Committee to make a decision when you're (apparently, maybe I'm
> misunderstanding) fine with asking ftp-master to reject such packages. Of
> those two things, I think the Technical Committee discussion is more open,
> collaborative, and accountable than asking ftp-master to reject packages!
> (Indeed, if I were ftp-master, I would just bump the request to the
> Technical Committee anyway, since I wouldn't want to make that sort of
> decision on the basis of just the ftp-master delegation authority.)
The root problem is that AFAICS there's been at least three different
ways the interested parties have expressed a desire they wanted this
getting enforced, for quite different reasons, that have gotten mingled
* Ban this everywhere at the dpkg level (dgit and Ubuntu interest).
* Ban this for all Debian at the ftp-master/lintian level via policy
(Ubuntu interest, fallback dgit interest).
* Ban this at the vendor level (Ubuntu interest)
Ideally? Yes, I'd like for any of the interested parties to get the users
of the feature and possibly convince them they are doing it wrong, and
if so why, and what they could do instead, or for the complaining part
to realize this might have some merit, and solve the apparent problems
they have with technical means.
But for the Ubuntu case, which is what I was explicitly discussing with
Steve, the reality is slightly different. The Ubuntu organization is
reigned by their own rules, and it's a different entity to Debian, and
how they reach their conclusions and policies is for them to decide, and
even if how they do that might not be my cup of tea, I'm just not involved
and I don't think it's my place to question their processes.
So, if Ubuntu, as an organization, decides that the ubuntu.series file is
not good for them, and they are going to ban it no matter what, probably
at their boundaries. Asking ftp-masters or lintian maintainers to honor
that wish and do that for them would just help them, and not change the
outcome much. Those two teams could obviously decide they do not want
to be in the middle of possibly angry maintainers and the derivative,
but that's for these teams to decide. I'm not sure why we'd reject that
at the Debian level TBH. It's also in their namespace so it seems only
fair they get to have a say IMO.