Re: [Curdle] Some work for the group

[Phillip Hallam-Baker <[e-mail]>]

It looks to me as if in effect the prehash vs pure issue is going to settle
out into a situation where all signatures are pure signatures at the
specification level with an additional hash being introduced in the
packaging layer where bulk data is involved.

I have implemented most of the popular packaging formats and most of my API
calls internally turn out to be for a 'sign and digest' operation rather
than Sign-this-digest. Most cases the bulk data is being put through a
digest that is then put into some other structure for signature. There is
almost always some sort of metadata you want in there in addition to the
content.

Another way to look at this is that Ed255x is a drop in replacement for RSA
Signature that removes the restriction on the size of the input data to be
signed but not the requirement for it to be buffered. Most packaging
formats end up being a signature over a small chunk of static data which in
the case of streamed content are virtually always reference to external
data.