On Thu, 28 May 1998, Jon Callas wrote:
Hal writes:
Jon - Unfortunately the signature packet retains a vestige of two
byte lengths:
...
It was a big mistake to do this but at the time we weren't anticipating
large signatures. Despite several people reviewing the design no one
caught this.
The only solution I see is to define a V5 signature packet format.
Hal
Actually I don't think it is a mistake. Signature packets shouldn't be
allowed to grow to infinity by being stuffed with data that isn't really
part of the signature information. Maybe a new literal type (I already
suggested a nested literal elsewhere), 's' to join 't' and 'b' that could
hold the information that is not part of the signature, but might be part
of a meta-certificate.
64K hashed and 64K unhashed is enough for any description. If something
has lots of little stuff exceeding 64k, just make more signature packets,
with only the mandatory packets in common.
Is it only me who thinks a signature packet should be a signature and not
a document, certificate, or whatever with signature?
--- reply to tzeruch - at - ceddec - dot - com ---