On Wed, Sep 25, 2002 at 07:04:07PM -0700, James A. Donald wrote: > This is perhaps a bunch of dumb questions, I am trying to play with > the library to get the hang of it. > > The elliptic curve benchmark code, as written, does not seem to test > the SEC2 recommended elliptic curves.
Yes, the benchmark code needs to be updated. > The ECP class does not seem to have any direct way of loading a SEC2 > recommended curve by name. This seems odd to me. What is the > rationale for this? That's because the a SEC2 curve include more than just a curve. It includes a base point and its order. The class that corresponds to a SEC2 curve is ECParameters. From ECParameters you can get to an ECP object by calling GetCurve(). > One reason why this might be a logical way of doing things is that > there might be some template class that takes either ECP or EC2N as a > parameter, and does load a SEC2 recommended curve by name and > provides a derived class that inherits directly from ECP, with only > knowledge of SEC2 added, but if that is the rationale, it is not > immediately apparent to me what this class is Note sure what you mean here. If my above answer does not answer your question, please ask again. > I am also having trouble grasping how one class connects to another > to provide an implementation of an abelian group. The things that > know how to do the operations of an abelian group seem to live inside > things that know how to use the properties of an abelian group to do > cryptographic operations. Are there classes are that know about > abelian group operations, but not cryptography? What do you mean? The classes that know about group operations are in ecp.h/cpp and ec2n.h/cpp. The class that know about cryptography are in eccrypto.h/cpp. I thought the seperation is pretty clear.
