This, I agree, is the worst part of RFC2440, causing me a lot of trouble (I
would actualy go further: it's almost useless in its present form; I ended
up figuring things out from PGP and GPG sources and interoperability
experiments). Instead of just tweaking the language, however, I would
suggest the inclusion of an overview table in which one can look up which
signature type is calculated on exactly what data, in what format, in which
order. It would be tremendously useful for implementers.
--
Daniel