Why different keys for -testing and non-testing?

Casey Dahlin cdahlin at redhat.com
Sat Jan 17 05:38:31 UTC 2009

Jesse Keating wrote:
> On Fri, 2009-01-16 at 22:23 -0500, Casey Dahlin wrote:
>>  I think its wrong to get the latter out of the keys (though the right 
>> way might mean touching rpm in a way we aren't allowed/able to). Once we 
>> have "This package came from Fedora" then for the rest of the info, we 
>> can just state it in a package header. If the headers are signed then we 
>> have the necessary level of security. We only need one key to provide 
>> the non-refutability. The rest of the information can just be stated.
> I'd rather state that in the repodata, rather than the rpm itself.
> Stating it in the rpm would mean changing the rpm file between -testing
> and updates, which would break the ability to hardlink, and would mean
> unnecessary churn.
That works too. The point is we shouldn't be using keys to categorize 
things. Only to authenticate them. And the only reason to have more than 
one key is to limit the effects of one of the keys being compromised 
(the all-your-eggs-in-one-basket issue).


More information about the fedora-devel-list mailing list