Gentoo Archives: gentoo-project

From: "Michał Górny" <mgorny@g.o>
To: gentoo-project@l.g.o
Subject: Re: [gentoo-project] [RFC] Asynchronous Council
Date: Fri, 26 Jun 2020 18:49:03
Message-Id: 4aeb190ba2808d4127d98b5ef3cfaeb7863d324d.camel@gentoo.org
In Reply to: Re: [gentoo-project] [RFC] Asynchronous Council by Ulrich Mueller
1 On Fri, 2020-06-26 at 13:59 +0200, Ulrich Mueller wrote:
2 > > > > > > On Fri, 26 Jun 2020, Michał Górny wrote:
3 > > This is the idea I'd like to submit for the first meeting of the new
4 > > Council. I'd like to hear your feedback first.
5 > > TL;DR: Council requests should always be discussed publicly
6 > > on the mailing lists. Council members should vote on them via bugs when
7 > > the discussion finishes. Meeting should focus on loose discussion
8 > > without decision making.
9 >
10 > I hope this isn't another of your belated April 1st jokes. :)
11
12 The rule is at most one joke per year, I guess ;-). No, this is
13 a serious thought.
14
15 >
16 > > The problem
17 > > ===========
18 > > Currently, most of the Council decisions are made during the monthly
19 > > meetings. This causes a few problems.
20 > > Firstly, it is hard to 'fit' the public discussion just for the meeting.
21 > > If it doesn't conclude before the meeting, Council is either operating
22 > > with partial feedback or defers the matter another month. If it
23 > > concludes earlier, things get forgotten.
24 > > Secondly, sometimes new arguments are brought during the Council
25 > > meeting. This often means that the Council has either to defer
26 > > the matter further to give people a chance to update their feedback,
27 > > or make decisions without giving people not present at the meeting to
28 > > reply.
29 > > Thirdly, this means that some people end up being inconvenienced to
30 > > appear at the meetings to support their cause. These meetings may
31 > > easily end up very late for some people, and it doesn't help when they
32 > > end up being very long because of the ongoing discussion.
33 >
34 > Could you substantiate these claims by some examples, preferably from
35 > a recent Council period?
36
37 The usual meeting time collided with my 'family time'. But I was
38 thinking more of Japan, somewhat in reference to the bigger problem
39 Trustees have. We're running a world-wide project, and we can't expect
40 to be able to manage to satisfy everyone, so we're effectively
41 discouraging some part of the community from taking active part in this.
42
43 >
44 > > Proposed solution
45 > > =================
46 > > Roughly, the proposal would be that:
47 > > 1. All public matters are discussed on the appropriate mailing lists.
48 > > 2. A request to the Council is made via filing a bug. If the request is
49 > > suitable for public debate, the discussion is linked in the bug.
50 > > Otherwise, appropriate restrictions are set on it and the discussion is
51 > > carried on the bug.
52 > > 3. Council members participate in the discussion, possibly acting
53 > > as moderators as well (i.e. asking people for clarifications,
54 > > encouraging them to express their opinions, keeping the discussion
55 > > on topic).
56 > > 4. If necessary, Council members may also discuss the matters via IRC,
57 > > possibly during ad-hoc meetings. However, the result of these
58 > > discussions must be brought to the mailing lists for feedback.
59 >
60 > So, who would be responsible for recording the log and writing a summary
61 > for these ad-hoc meetings? I see it as a huge advantage of the monthly
62 > meetings that the schedule is well structured and everything is in the
63 > log and summary.
64
65 I don't think it's important to set this in stone right now. A few
66 possibilities: whoever called for it, a person chosen at the beginning
67 of the meeting, the chair for the month that the meeting takes place
68 in...
69
70 > Plus, regular meetings are currently being announced some time in
71 > advance, so developers can take notice and participate.
72 >
73 > > 5. When the discussion settles (either reaching a conclusion or
74 > > saturating all the arguments), the Council votes on the bug.
75 >
76 > As a matter of fact, that hasn't really worked so well in the past.
77 > It sometimes took quite a while until all votes were collected.
78
79 I believe this would change if this became the primary workflow.
80
81 An extra benefit is that it would make sure that Council members
82 participate more often than one hour a month.
83
84 > > 6. Decisions are announced on the -dev-announce mailing lists.
85 > > If multiple votes are carried simultaneously, multiple decisions can be
86 > > announced in one mail.
87 > > What about the meetings?
88 > > ========================
89 > > The Council can continue to meet monthly with the two common agenda
90 > > items: reiteration of open bugs and open floor. The former can serve
91 > > as a synchronization mechanism for other requests. The latter provides
92 > > the community an opportunity to find all of the Council members
93 > > available at the same time.
94 >
95 > Nope, the Council *must* continue to have at least one monthly meeting,
96 > because GLEP 39 mandates it:
97 > - The council must hold an open meeting at least once per month.
98 > - Council decisions are by majority vote of those who show up (or their
99 > proxies).
100
101 While I do know the complications of changing GLEP 39, if we want to
102 change workflow we should obviously account for the possibility of
103 changing it.
104
105 >
106 > IMHO it would also be against the spirit of GLEP 39 if most Council
107 > decisions were to take place outside of monthly meetings.
108 >
109 > > How does that improve things?
110 > > =============================
111 > > Most importantly, all the discussion is carried off in public.
112 >
113 > Council meetings are public already, so no difference there.
114
115 I'm sorry, 'public' does not carry my meaning correctly. My point is,
116 mailing list discussion provides a better opportunity for people
117 (possibly having little time, restricted schedule, mismatched timezone,
118 needing to do more research) to participate than a meeting that requires
119 participation at a particular time and reasonably fast answers.
120
121 > > The privileged position of people attending Council meetings is removed.
122 > > Decisions aren't made in the middle of public discussion, or long after
123 > > it settled. Council members bring their arguments to the debate
124 > > earlier.
125 > > Since the Council is no longer bound by the monthly cycle, decisions can
126 > > be made faster if there is overall agreement.
127 >
128 > If there's overall agreement (e.g., consensus in gentoo-dev), you
129 > normally don't even need a decision by the Council.
130
131 GLEPs are one thing I was thinking of. Even if the GLEP is clear-cut
132 and nobody raises any concerns, we have to wait for the meeting or call
133 for non-standard bug vote.
134
135 > > The meetings become shorter, on one hand saving Council members' time,
136 > > on the other making it possible for developers to participate in open
137 > > floor without waiting for all agenda items to be processed.
138 >
139 > That would be easy to solve by putting open floor first on the agenda.
140 >
141
142 I don't think that would be desirable when we expect to have on-topic
143 discussion on the agenda.
144
145 > > WDYT?
146 >
147 > Overall, this sounds like replacing a well structured workflow by
148 > something that would be more irregular and intransparent. Especially,
149 > discussions would be scattered over several bugs and mailing list
150 > postings, instead of having them in one place.
151 >
152
153 They already are. Actually, they would less scattered as the IRC would
154 no longer be present in primary discussion flow (in favor
155 of the original ml discussion).
156
157 --
158 Best regards,
159 Michał Górny

Attachments

File name MIME type
signature.asc application/pgp-signature