Re: [jose] thoughts on header criticality from recent polls

["Manger, James H" <[e-mail]>]

A problem with both the FLAT and STRUCTURED syntaxes is that there are lots of non-critical fields.

From JWE (draft 8): critical fields are “alg”, “zip” – and I think that is it. “jku”, “jwk”, “x5u”, “x5t”, “x5c”, “kid”, “typ”(?), and “cty” are non-critical. Even “apu”, “apv”, “epu”, “epv”, “enc”, and “epk” can probably be marked as non-critical (you might not be able to successfully process a JOSE message that has these fields if you ignore them, but you will not *misinterpret* the JOSE message which is the only real concern).

From JWS (draft 8): the only critical field is “alg”.

I doubt anyone wants most field names duplicated in an “ign” array [FLAT syntax].

I don’t like the STRUCTURED syntax. The prospect of writing code that has to look in 2 places for every field it is interested in is fairly appalling (eg k = header.kid !== undefined ? header.kid: header.ign.kid). Not to mention the extra complexity/corner-cases/security-holes from the possibility that a field has 2 different values: one at the top-level, and another under “ign”.


Having 1 critical field that must be given a new value if a new message could otherwise be misinterpreted seems much simpler.


--
James Manger

From: [e-mail] [mailto:[e-mail]] On Behalf Of Mike Jones


FLAT SYNTAX:
    {"alg":"ES256",
     "ign":["notes"],
     "notes":"This header field can be safely ignored"
    }

STRUCTURED SYNTAX:
    {"alg":"ES256",
     "ign":{
       "notes":"This header field can be safely ignored"
      }
    }