The v5 key format contains a series of octet fields that are redundant with other parsing information contained in the packet.
These fields increase the size of the packet on the wire, and they create a new risks of incompatibilities: what if these fields don't match what the rest of the packet lengths suggest? how does an implementation deal with that kind of overflow?
However, they provide a useful feature, which is that they permit calculation of the fingerprint based on a secret key packet even for an unknown public key algorithm.
This change proposes removing these fields from the wire (making v5 public keys functionally equivalent to v4 public keys, but with a different fingerprint calculation method), while adding one trailing field to the v5 secret key packet to facilitate secret key to public key transformations (and therefore fingerprint calculation).
Closes: #43 (closed)