1 |
Mike Auty wrote: |
2 |
> Ciaran McCreesh wrote: |
3 |
> | |
4 |
> | Signing offers no protection against a malicious developer. |
5 |
> | |
6 |
> |
7 |
> I had envisaged a system whereby when the tree was synced, as was some |
8 |
> kind of master signed list of all acceptable dev-keys. Every package |
9 |
> would also be signed, and would only be installed when signed. As soon |
10 |
> as a dev becomes a liability their key is removed from the list/revoked. |
11 |
> ~ On next sync any packages or package upgrades signed after the time of |
12 |
> revocation would not be installed. There would be a window of |
13 |
> vulnerability, but no bigger than with revoking a dev's access to the |
14 |
> tree. Do you think this would offer suitable protection for users from |
15 |
> a malicious dev or not? |
16 |
There has been some previous work which has never been finalized, for |
17 |
all interested parties: |
18 |
http://viewcvs.gentoo.org/viewcvs.py/gentoo/users/robbat2/tree-signing-gleps/ |
19 |
|
20 |
Getting this cleaned up and ready for discussion would be quite valuable. |
21 |
> |
22 |
> I understand there are difficulties with eclasses, etc, which is why the |
23 |
> current implementation is still not widely used or mandated, but I'm |
24 |
> more interested in the feasibility of the idea. |
25 |
It can be done if people can agree to a policy and allow the |
26 |
programmatic and infrastructural changes to happen. |
27 |
|
28 |
Have fun, |
29 |
|
30 |
Patrick |
31 |
-- |
32 |
gentoo-dev@l.g.o mailing list |