3 ms·
tptacek was saying everyone should be cautious about PKI-based JWTs and provided encrypted cookies as an example of something that is rarely (if ever) implement
by giaour 9y ago
tptacek was saying everyone should be cautious about PKI-based JWTs and provided encrypted cookies as an example of something that is rarely (if ever) implemented safely and securely with PKI cryptography. My point is that while the 'seal' function of PAST may be of questionable value, the 'sign' function is not. I think it's important to disambiguate those two uses of asymmetric cryptography.
- lvh 9y agoGotcha, thanks! That makes more sense now. I agree that they're worth examining separately, but the more general point still stands that there is a lot more that can go wrong with sign than with aead. The arguments for broken RSA-encrypted cookies apply similarly to broken signed cookies. It's not clear to me for example that an invalid curve attack against seal is intrinsically worse than a nonce (ab|re)use bug against ECDSA.