1 |
I've picked the top mail in this thread to response to a single issue, |
2 |
because I think it's important. Dilfridge was in #gentoo-infra earlier |
3 |
this week, trying to figure out projects->alias generation, and we had |
4 |
a productive discussion of some the fundamental issues behind that goal. |
5 |
|
6 |
This email is a result of that discussion. |
7 |
|
8 |
On Wed, Sep 30, 2015 at 08:15:37PM +0200, Michał Górny wrote: |
9 |
> 1c. We may also automatically add members to mail aliases from |
10 |
> herds.xml if someone really wants that. But again, not a priority. |
11 |
|
12 |
A lot of this discussion seems to conflate project membership with being |
13 |
on a mail alias, and this is problematic in various ways. |
14 |
|
15 |
0. Some projects have MORE than one alias. |
16 |
|
17 |
1. While project membership MUST be public, it's entirely possible that |
18 |
some or all of the alias members are private: |
19 |
1.1. Infra & Security aliases: Some of the aliases themselves might be |
20 |
secret, used to filter some mail. The list of addresses on the alias is |
21 |
also private. |
22 |
1.1.1. "private alias" => members of the alias are hidden, but the alias name is published |
23 |
1.1.2. "secret alias" => the alias name is not published |
24 |
1.2. Some contributors to projects LIKE their privacy and/or have |
25 |
specific addresses for the alias. This set includes upstream devs that |
26 |
care about Gentoo bugs for their app. |
27 |
|
28 |
2. Just because a developer is a member of a project, does not always |
29 |
they want ALL the mail on a given alias. Some aliases, esp the older |
30 |
ones or more central ones are firehoses of both real bug mail and |
31 |
spam. |
32 |
|
33 |
The combination of these points has a few implications: |
34 |
|
35 |
A. Developers need to be able to opt-out of an alias while retaining |
36 |
project membership. |
37 |
A.1. Should this fact be visible? |
38 |
|
39 |
B. Non-developer members of a project need to addable to an alias by a |
40 |
developer |
41 |
B.1. While preserving the privacy of that contributor entirely, if they |
42 |
don't want to show up on a listing of members) |
43 |
B.2. While preserving the privacy of the contributor's email address, if |
44 |
they don't want their email address published. |
45 |
B.2.1. Additional implication is that the email addresses CANNOT be in a |
46 |
public repository at all! |
47 |
B.3. This scares some developers, not known that a non-developer is on a |
48 |
mail alias. |
49 |
|
50 |
From this, I have a rough plot of a data model that tries to take it |
51 |
into account. It doesn't cover where to store it, other than it needs to |
52 |
be private. |
53 |
|
54 |
= Projects have members |
55 |
== members fall into two groups: developers, contributors |
56 |
== members can be public or private, this controls if they are shown on the wiki etc. |
57 |
== developers are public members by default |
58 |
== contributors are private members by default |
59 |
= Projects have mail aliases |
60 |
== members opt into project mail aliases. |
61 |
== Public project members NOT in a mail alias are visibly flagged. |
62 |
== Private project members IN a mail alias are visibly flagged (how to |
63 |
do this without exposing them too much?) |
64 |
|
65 |
I think part of the solution will lie in separating the identity of a |
66 |
contributor vs their email address: Allow viewing of who is on an alias |
67 |
without exposing the actual email address. |
68 |
|
69 |
"John X. (upstream author)" in a alias listing is MUCH more useful than |
70 |
a random email address. |
71 |
|
72 |
-- |
73 |
Robin Hugh Johnson |
74 |
Gentoo Linux: Developer, Infrastructure Lead |
75 |
E-Mail : robbat2@g.o |
76 |
GnuPG FP : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85 |