1 |
On Fri, 2006-08-25 at 19:13 +0200, Wernfried Haas wrote: |
2 |
> On Thu, Aug 24, 2006 at 06:29:03PM -0400, Chris Gianelloni wrote: |
3 |
> > Quite frankly, I think that with a properly run community, there should |
4 |
> > be no need for a "Developer Relations" project, since it should be |
5 |
> > mostly self-policing. |
6 |
> |
7 |
> With 300+ people, i severely doubt self policing would work. I assume |
8 |
> you were mostly thinking about conflict resolution when you wrote |
9 |
> this, but there are other things like |
10 |
|
11 |
You assumed wrong. While I was mostly thinking about conflict |
12 |
resolution, I still don't think that in a properly run project there is |
13 |
a need for an "HR" project. |
14 |
|
15 |
> - recruitment |
16 |
|
17 |
This goes into the whole thing about bringing people into the project, |
18 |
as well as having a longer probation period. The mentor should really |
19 |
be responsible for the developer. If things were working smoothly, |
20 |
there really shouldn't need to be much work done for recruitment. |
21 |
Remember, we're talking a meritocracy here. We shouldn't just be |
22 |
bringing on some guy because he says he'll do something. We should be |
23 |
bringing on people that have *proven* that they can and will do |
24 |
something. |
25 |
|
26 |
> - retirement of developers that |
27 |
> - quit |
28 |
> - go AWOL |
29 |
|
30 |
See, you missed that we're talking with the idea of people belonging to |
31 |
a project. If you work on my project and quit, I'll know. If you go |
32 |
AWOL, I'll know. I can then simply ask Infra to remove your access. It |
33 |
really should be that simple. If Infra is unable to do so due to being |
34 |
understaffed, then they should get more staff. |
35 |
|
36 |
> etc. |
37 |
> |
38 |
> In an ideal world with a self-policing community this could work out, |
39 |
> however i rather tend to assume one of these things happen in the real |
40 |
> world: |
41 |
> - People get recruited by anyone in whatever way and have no idea about |
42 |
> our policies, breaking the tree in their first commit. |
43 |
|
44 |
Uhh... like this hasn't happened in the past? |
45 |
|
46 |
> - Developers retire and no one removes their access due to lack of |
47 |
> procedure |
48 |
|
49 |
If the developer belongs to a project, the manager knows it and asks for |
50 |
their removal. Is there need for any more procedure than that? |
51 |
|
52 |
> - Devs go AWOL, no one notices. If someone notices, perhaps he starts |
53 |
> volunteering doing this kind of clean-up work, and technically a new |
54 |
> devrel project emerges. |
55 |
|
56 |
See, you seem to be assuming that I haven't thought this through. It |
57 |
would be impossible for a developer to go AWOL without their |
58 |
manager/lead/whatever you want to call them noticing if everyone were a |
59 |
member of a project. This doesn't mean you can only be on one project, |
60 |
it means you must be on *at least* one project. No more projects, no |
61 |
longer a developer. It's simple, yet effective. |
62 |
|
63 |
> > Beyond that, the leadership should have the power |
64 |
> > and the ability to take care of problems in a timely manner without the |
65 |
> > need for droves of bureaucratic process. |
66 |
> |
67 |
> The process (http://dev.gentoo.org/~plasmaroo/policy.xml) is |
68 |
> reorganized and should fulfill both your concern for a timely manner |
69 |
> and is much less bureaucratic. |
70 |
|
71 |
It does not fulfill my concerns in any way. Developer Relations is not |
72 |
Gentoo's leadership. |
73 |
|
74 |
> Also, there's a lot of stuff happening other than the conflict |
75 |
> resolution stuff with regard to ombudsman and often kloeri resolving |
76 |
> things - you don't read that on the news, but i'm not sure if council |
77 |
> members should spend a lot of their time to resolve silly conflicts |
78 |
> between devs, they're elected make decisions, not obsolete devrel. ;-) |
79 |
|
80 |
With a proper structure, the council wouldn't need to be concerned with |
81 |
this sort of thing. Here's an oversimplified example. |
82 |
|
83 |
You are in projects A, B, and C. You haven't done anything for project |
84 |
A in six months, and the manager has noticed, and removed you from his |
85 |
project. He looks to see if you are in other projects, which you are, |
86 |
so he is done. He *could* email the other project leads at this point |
87 |
to see if you're still active, but it wouldn't be required. Each |
88 |
project maintains itself, after all. Project B has done the same since |
89 |
you haven't done anything for them in 3 months. Project C's manager |
90 |
notices that you haven't done anything in 2 months, with no word that |
91 |
you were leaving. He also notices that you aren't in any other |
92 |
projects, so he reopens your dev bug and asks infra to retire you. |
93 |
Process completed and no council involvement, whatsoever. |
94 |
|
95 |
The same sort of process really should be used for conflict resolution. |
96 |
Hell, at least our current policies dictate this, but it never happens, |
97 |
in practice. Instead, everybody goes directly to devrel. If two |
98 |
developers within a single project have a conflict. It goes to that |
99 |
project's manager. If it is between two devs in two projects, it goes |
100 |
to those two managers. The only reason it would need council |
101 |
involvement is if the managers cannot make a decision themselves, or |
102 |
possibly on appeal. There's really no other reason for council |
103 |
involvement. |
104 |
|
105 |
> Btw, the new policy also includes the possibility of refering a |
106 |
> decision to the council in certain cases, see "Resolution and Appeal". |
107 |
|
108 |
I've read the policy. |
109 |
|
110 |
> > I'm sure nearly every member |
111 |
> > of devrel would agree that they would love to see a Gentoo where devrel |
112 |
> > simply wasn't needed. |
113 |
> |
114 |
> I assume you're only refering to conflict resolution again, and i |
115 |
> agree it would be great. I just don't think this is ever going to |
116 |
> happen as long there are more than 50 developers. |
117 |
|
118 |
Quit assuming I mean anything, you're batting zero for two right now. |
119 |
|
120 |
Luckily, I wasn't asking if you thought it was possible. I've merely |
121 |
been stating that it should be possible. There are countless projects |
122 |
out there, many with many more developers than Gentoo, that are capable |
123 |
of maintaining themselves quite well. Why are we so different? |
124 |
|
125 |
-- |
126 |
Chris Gianelloni |
127 |
Release Engineering - Strategic Lead |
128 |
x86 Architecture Team |
129 |
Games - Developer |
130 |
Gentoo Linux |