3 ms·
One of the good things in TLS 1.3 was that cryptographers were brought in early, the TLS 1.3 core design is a product of cryptographers plus engineers, and then
by tialaramex 8y ago
One of the good things in TLS 1.3 was that cryptographers were brought in early, the TLS 1.3 core design is a product of cryptographers plus engineers, and then the work to file off sharp corners and make it work in the real world are the engineering problem, whereas previous versions the engineers built something that seems to work, then (sometimes years later) the cryptographers were shown the results and usually they are not entirely comfortable but it's too late.
Maybe SNI needs (if it hasn't had already) a few rounds of smart crypto postgrads and postdocs trying to figure out what is possible here. If the answer is that nothing we can possibly do fulfils all the criteria, well, that's sad but we must find a compromise knowing this limitation, if there's a way to satisfy all requirements in this document, maybe they can find it. As it is today, it feels a bit hopeless, or at least I'm worried that it does.