Re: [quicwg/base-drafts] Add an appendix containing test vectors for "Initial". (#1573)

[Christian Huitema <[e-mail]>]

Yes, it may well be that including a "real" packet as a test vector is a bad idea. If we do that, then we will need to update the version number (bytes 2 to 5) from 0xff, 0x00, 0x00, 0x0d to 0xff, 0x00, 0x00, 0x0E for draft 14, and so on for each draft. This does not affect the computation of AEAD and PNE keys, as they are only a function of the salt and connection ID, but it does affect the sample used for PNE in the example.

Suppose that instead of a full packet we had a truncated packet, with just enough bytes to get the PNE sample. We will lose the capability of doing a full test "up to packet decoding", but in my experience that's not a very big deal. Implementations can use different unit tests for verifying the implementation of AEAD. But for that loss we get a big gain: we can update the test vector by just flipping the version bits, without having to worry about the rest.

Unless we change the salt or the algorithm of course. But then if we do that we know that we have to change the whole test vector no matter what.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/pull/1573#issuecomment-413425717