1 |
On Fri, Nov 12, 2004 at 11:30:08AM -0500, Chris Frey wrote: |
2 |
> While I'm not opposed to ssl/ssh links in any way, I think this would be |
3 |
> more work to install than the signature method that already has a patch. |
4 |
> |
5 |
> Consider: |
6 |
> |
7 |
> Patch method: |
8 |
> - no mirror needs to be updated |
9 |
> - users can continue to use any available mirrors for |
10 |
> the webrsync tar (do they exist?) |
11 |
> - the main gentoo server only has to serve the signature |
12 |
> (this could be put on a single mirror too, point being |
13 |
> that the signature doesn't have to be on every mirror |
14 |
> to be effective) |
15 |
> |
16 |
> SSL/SSH method: |
17 |
> - either every mirror needs to support it |
18 |
> - or anyone who is concerned, suddenly stops using mirrors |
19 |
> and switches to the main server |
20 |
> - doesn't detect cases where a mirror is compromised |
21 |
> |
22 |
> Just points to be aware of when considering SSL/SSH. |
23 |
FullACK |
24 |
|
25 |
even worse, it would stress, the CPU even more, and so on. |
26 |
|
27 |
these "solutions" do not interfere with each other but |
28 |
if these are combined I can't see a great benefit |
29 |
either. |
30 |
|
31 |
the only benefit is, that a man in the middle would not |
32 |
be able to guess the state, my local system is in (eg: if |
33 |
I am fetching an actual .ebuild file of portage, the mim |
34 |
can be sure that the local version is not the actual one) |
35 |
not a big benefit. |
36 |
|
37 |
anyway, as someone seems to use some sort of that "solution" |
38 |
I got curious and wanted to get some more details... |
39 |
|
40 |
regards, |
41 |
|
42 |
klaus |
43 |
|
44 |
-- |
45 |
gentoo-security@g.o mailing list |