| 1 |
I searched -dev, -user, and bugzilla but found nothing about signing |
| 2 |
packages and keeping sums or some other verification scheme withing the |
| 3 |
ebuild script for Portage. |
| 4 |
|
| 5 |
Should I/we consider a RFE for future versions that would (among other |
| 6 |
things I'm sure): |
| 7 |
|
| 8 |
- Sign the ebuild scripts w/ a Gentoo key that a few key players would |
| 9 |
have access to. Alternatively, for packages not 'core' to the system |
| 10 |
they could be signed by the author. Emerge would need a way to manage |
| 11 |
this through GPG but we can work on that. ( And you'd accept new keys |
| 12 |
manually, have them downloaded from MIT or keyserver.net, etc. ) |
| 13 |
|
| 14 |
- Embed in the signed ebuild script the md5sums of all the tarballs and |
| 15 |
patches you're about to go snag for building. |
| 16 |
|
| 17 |
This would help w/ security of packages on mirrors and such as well. |
| 18 |
|
| 19 |
I have a sneaking suspicion somebody will say something similar is |
| 20 |
already done, it isn't needed, etc. For some reason I would think this |
| 21 |
would already be covered in design goals. |
| 22 |
|
| 23 |
<shrug> |
| 24 |
|
| 25 |
Just a thought... -Ali |
| 26 |
|
| 27 |
-- |
| 28 |
OpenPGP key 53F7FF5F |
| 29 |
-- |
| 30 |
Bear in mind that, in 30 years' time, all that might remain of the |
| 31 |
system you're building today is the memory of its more newsworthy |
| 32 |
security failures. -- 'Security Engineering', Ross Anderson |
Archives