Boost logo

Boost :

From: Gerhard Wesp (gwesp_at_[hidden])
Date: 2001-05-03 03:43:57


> I'm not sure whether we want to have a singular implementation into
> boost without thinking about the interface issues, e.g. how to
> concatenate the execution of two algorithms, and have a general
> design for the whole application domain.

Discussing the interface issues would be one of the interesting parts of
this issue!

> Anyway, we may as well start with one concrete implementation of
> an algorithm and expand on that later. We did that with Beman's

This also was my general direction of thought when I made the proposal.
Different block ciphers could be added, then classes to wrap block
ciphers into stream ciphers, and so on.

> Oh, btw, the interface should not be std::vector<byte>-based, but
> be iterator-based instead.

For a block cipher? Hmmm... Well, probably you're right, but I'd argue
to restrict ourselves to random access iterators in that case (to be
able to check the block size).

-Gerhard
PS: what about export regulations in certain countries? Would adding
cryptography create problems?


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk