1 |
On Tue, 2006-01-10 at 23:57 -0500, Andrew Muraco wrote: |
2 |
> What I meant to say is, having this alternative tree method (as |
3 |
> described here) would mean that portage would handle everything the |
4 |
> exact same as it already does, which means that if someother tree was |
5 |
> accidently sync'd or replaced the local one, portage would react and |
6 |
> want to update everything, because its not 'aware' that there is a |
7 |
> difference in the first place. (now that I think about it, how likely is |
8 |
> it that something like that will happen?) |
9 |
|
10 |
So someone would "accidentally" change the SYNC variable, then |
11 |
"accidentally" run "emerge --sync"? That seems rather unlikely. |
12 |
|
13 |
> The method described here would also open up the oppurtunity for "fake" |
14 |
> enterprise trees, without having something to double check that the tree |
15 |
> that we have is indeed the one that is wanted, it would be possible for |
16 |
> a hacked rsync server (or a bogus one) to host the enterprise (stable) |
17 |
> trees with extra ebuilds which could compromise security (/me thinks of |
18 |
> emails warning about Microsoft's patchs and links which point to |
19 |
> infectious websites.) |
20 |
|
21 |
What exactly makes the release trees any more vulnerable to this than |
22 |
current portage? |
23 |
|
24 |
> Maybe this is something thats not very likely to happen, but it still is |
25 |
> important to note that an enterprise tree has to be secure. |
26 |
|
27 |
You will notice that I *never* call it an "enterprise" tree, so don't |
28 |
make that mistake yourself. |
29 |
|
30 |
-- |
31 |
Chris Gianelloni |
32 |
Release Engineering - Strategic Lead |
33 |
x86 Architecture Team |
34 |
Games - Developer |
35 |
Gentoo Linux |