1 |
On Wed, Jan 4, 2017 at 12:58 PM, Kristian Fiskerstrand <k_f@g.o> wrote: |
2 |
> With increasing focus on security in various contexts I'd like to |
3 |
> propose that we start discussing catching up with other distributions |
4 |
> and start requiring new developers' OpenPGP keyblocks to have at least |
5 |
> two signatures from existing developers before applications can be |
6 |
> made[A]. Amongst other things This helps building the Gentoo Web of Trust. |
7 |
> |
8 |
> |
9 |
> E.g [Debian] has the following requirement: "To maintain the strong Web |
10 |
> of Trust that connects all Debian Developers, Applicants need to |
11 |
> identify themselves by providing an OpenPGP key that is signed by at |
12 |
> least two official Developers. To further ensure their identity, |
13 |
> signatures by other people (who do not need to be DDs, but should be |
14 |
> well connected in the overall Web of Trust) are strongly recommended." |
15 |
> |
16 |
|
17 |
Looking at our developer map this seems incredibly impractical. I |
18 |
know I've yet to actually bump into another Gentoo developer. Sure, I |
19 |
could fly out to SCALE or FOSDEM (which are about equidistant), but |
20 |
this seems a bit much for a requirement, even if I'm likely to get |
21 |
around to it one of these years. |
22 |
|
23 |
Also, we have fairly specific requirements for our gpg signing keys, |
24 |
so there is a good chance that any existing keys that candidates have |
25 |
which bear signatures may not be usable for Gentoo, meaning that |
26 |
they're starting out from ground zero. I don't know if the intent is |
27 |
that the signatures come from keys that meet our gpg key requirements, |
28 |
but if so that will mean that most candidates will not have an |
29 |
existing web of trust either. Personally I met the Gentoo gpg |
30 |
requirements by just generating a new key used for Gentoo signing |
31 |
only, and it has no signatures at all. |
32 |
|
33 |
Sure, it makes sense in an ideal world, but if we're going to go along |
34 |
this route I think we need to come up with a more practical way of |
35 |
getting developer signatures than bumping into them at conferences, or |
36 |
happening to live nearby one. I'm surprised Debian is able to make it |
37 |
work, even with their larger developer counts. I guess it could work |
38 |
in areas with high concentrations, like Silicon Valley. Maybe you |
39 |
could get by with video conferencing and holding up passports/IDs, |
40 |
though good luck finding a client for that which meets our social |
41 |
contract and works on a Chromebook. :) |
42 |
|
43 |
-- |
44 |
Rich |