On Fri 2015-05-08 14:56:13 -0400, intrigeri wrote:
> I trust SSH remote host authentication much more than I trust whatever
> library + certificate authorities bundle Git is using for HTTPS.
Understood :) I just don't like authorizing use of my secret key
material in cases where i don't need to, so i usually aim to minimize
that.
In the case of a dev sending code back to a repo, the developer's push
should fail as a non-fast-forward push, if their view of the repository
had been tampered with by a network attacker capable of modifying HTTPS
but not SSH (i haven't actually tested this scenario, fwiw).
And of course OpenPGP verification of git commits and tags should be
possible anyway, regardless of the choice of transport.
Anyway, those are my reasons for suggesting the change, but i totally
understand your reason for not taking the approach too.
> If others like the proposed documentation change, feel free to apply
> it to Git, but I personally won't implement what it suggests (which is
> no big deal, of course :)
I don't think i have commit privileges to the wiki directly, and it
probably shouldn't be my call on how tails as a whole advises the
project anyway.
And i'm ok if the project decides to reject these suggestions, i just
wanted to offer them.
Regards,
--dkg
