Ian Grigg wrote:
Why can't a self-signed cert/key revoke itself?
How would it do so? Would it publish a CRL listing itself? And if you found a CRL that listed its signer's cert, would you trust that CRL?
In the original scenario, Ben was leaning towards person to person communications, such as email. So, to do a revocation, the user could hit the button to revoke (which might create a CRL if that is the best way to do it) and then mail the results to the people in the address book.
In browsing, one could publish the revocation, but as self-certs would be normally used for low monetary value, or otherwise protected activities, then just replace the self-signed cert with another and tell everyone you mucked up.
Isn't that like choosing whether or not to believe the person who says "everything I say is a lie"?
No, the key is saying that "I am compromised" and the key is as authoritive in its statements as anything else.
Even if this is a false statement (the owner only thinks it is true), it is still acceptable as a true statement. It simply means there are some cases where one is over-zealous.
Mind you, revocations seem rather rare.
Look at the size of any CA's CRL. Even cacert's CRL seems to have a lot of entries, and seems to have expanded at a significant rate.
Oh, ok! Now, how many of those are actual results of compromise? As opposed to routine replacements or expiries or other benign effects. Are we saying that CACert has a lot of compromises already? That would be a surprise.
Perhaps I should have said compromise revocations are rare, or important revocations are rare...
iang
_______________________________________________ mozilla-crypto mailing list [EMAIL PROTECTED] http://mail.mozilla.org/listinfo/mozilla-crypto
