Juju Delegated Team
robie.basak at ubuntu.com
Fri Sep 23 19:52:18 UTC 2016
I'm off next week. I'm sorry that I didn't realise before that I
wouldn't be able to attend the DMB meeting on Monday. I thought I could
leave my opinion and proxy vote here by email.
I'd like to thank you again for your work on the Juju packaging. Your
work is valuable to Ubuntu whether you can upload directly or not.
I think we can break the question down into five pieces as follows. I've
labelled them so that I can refer to them below. The questions are:
[A] Is it OK in principle to have packages in the archive that we
consider that only core devs should touch?
[B] If [A], does Juju qualify?
[C] If [A], then is it OK to still permit PPU from non core-devs on the
basis that they upload bugfixes only, and that any other type of upload
should be sponsored or reviewed by a core dev?
[D] If not [B], then should we give Nicholas PPU?
[E] Else if [C], then should we give Nicholas PPU?
If at the meeting these questions are voted on, then my votes are:
[A]: +1, [B]: +1, [C]: +1, provided that reviews are explicit through
signed uploads by the sponsor, [D]: -1, [E]: -1.
I feel that I owe objective reasons for any -1, so I've written them up
Having been close to Juju packaging in the past I feel that I am in a
good position to make a decision on how I should vote, and I don't
otherwise have a conflict of interest in this case. Thus I think it is
part of my responsibility that I provide a definite +1 or -1 on any
question on this matter. However any decision in this area is a very
subjective and difficult one to make, and I accept and would still be
happy if our conclusion as a body ends up being different. My opinion is
my vote; we use the other six members' opinions to form the board's
Nicholas asked in the IRC meeting if I was concerned that uploads would
go through that weren't acceptable. I think it is the point of the DMB
to be concerned about this; any doubt I have about granting upload
rights is by definition this concern.
I've not been on the DMB long, and feel that I am lacking in experience
in this type of DMB decision making. I welcome the opinions of others,
especially the TB and other senior Ubuntu develepors. I'd be happy for
my opinion to be changed by this group. Though of course you can always
appeal any decision to the TB, I encourage escalation to the TB for [A]
and [B] in particular, in part because I am keen to hear their opinion.
My reasoning for a +1 on [A] is that Ubuntu uploaders need to think of
the archive as a whole. I don't think it's sufficient to say that Juju
packages only affect Juju users. We want to present a consistent
experience for users across all packages, as well as internally in
packaging for Ubuntu developers to remain effective as a team. And
unlike snaps, packages interact with other packages on a shared system.
On [B], I think that Juju qualifies largely for the reasons I presented
in a previous email to this thread.
I think my concerns only apply to big picture packaging decisions, such
as how to deal with bundled libraries, static linking, golang/gccgo and
so on. Minor updates such as for upstream microreleases are clearly out
of scope of my concern. Therefore I feel that there is no reason to
hinder progress for this. It would be more pragmatic for the DMB to draw
a line and ask PPU uploaders not to cross it in this special case.
However, I want to require that sponsors for major changes sign and
upload themselves, so that there can be no confusion over what was
reviewed and acknowledged by whom. This is in part driven by my concern
over communication (more about that below).
Nicholas said in the IRC meeting: "I trust I've shown to be a good
steward. If I haven't, I don't deserve any rights. And if I have, why
shouldn't I have full trust?". But if we followed this principle to its
logical end, there would be no such thing as PPU or packagesets. Once
someone is an uploader the DMB's decisions would be enforced by uploader
trust only, and not by Launchpad ACLs. But we don't have it set up that
way, presumably for good reason. I am interested in others' views on
this, particularly from the TB. But without any other input, and given
that it is the DMB's sole remit to control upload access to the archive,
I feel that it is entirely appropriate for the DMB to request a
restriction in this manner.
[D] is automatically a -1 for the same reasons as [E].
[E] is the only question that applies to Nicholas personally, rather
than generally about uploads to Juju related packages. I am -1 for two
Based on your application, I see only three sponsored uploads for Juju,
and none for the related packages that you asked to be part of the same
packageset. The minimum figure I have in my head for PPU is somewhere
around five. The difference doesn't seem like much in absolute terms,
but it's a pretty big percentage. The work you've presented in your
application in this area seems pretty sparse to me. I appreciate that
you may have worked on other uploads and done other related work that is
not attributed to you, but I can only go on the basis of what I can see
on your application.
Note that the sponsorship miner seems to be down right now (timing out
on DB?). If that reveals more, I would reconsider this point.
I appreciate you bringing up the question of architecture support for
stable releases on the ubuntu-release list. That's a big positive
against the point I'm about to bring up.
I regret that we have not interacted much, even though I'm quite aware
of you being around and active in Ubuntu for quite a while. The two
interactions we did have on IRC last week, however, did not go
very well from my perspective. I didn't feel that you understood what I
was asking, and that your answers risked being quite misleading, even if
this was unintentional.
With the understood complications in Juju packaging, I think it is
essential that decisions in non-trivial uploads happen by real consensus
rather than by misunderstood consensus. The discussions we had give me
concern that this could go wrong.
This may be me, or it may be an impedence mismatch between us that
wouldn't happen with other Ubuntu developers. I trust that other DMB
members can decide for themselves, and for the outcome of the DMB vote
to decide whether it is just me or not.
Thank you for reading this far. I'd like to re-iterate that this is my
subjective opinion, that I welcome others on the DMB to disagree, and
would be happy if you wanted to take this matter to the TB if a DMB vote
does not go your way.
More information about the Devel-permissions