Re: Cryptography for Boost.

From: Vinnie Falco (vinnie.falco_at_[hidden])
Date: 2020-02-24 16:05:38

On Fri, Feb 21, 2020 at 8:09 AM Mikhail Komarov via Boost
<boost_at_[hidden]> wrote:
> ...wether there is still any interest in cryptography library for Boost and is
> there any sense to show the community what do I have for now?

I think there is utility in a cryptography _framework_ but not
necessarily a cryptography _library_. Implementing cryptographic
algorithms is something that is best left to experts. Generally, a new
implementation of an algorithm is worse than an old implementation
simply because the old implementation has had more provably-correct
hours in operation.

What I would like to see in a _framework_ is a set of named
requirements. Examples:

* BlockCipher
* MessageAuthenticationCode
* CompressionFunction
* OneToOneMapping

etc...

This is more of a written / documentation specification than C++ code,
although having some C++ metafunctions to check that types meet
syntactic requirements is nice.

With a set of named requirements in place, building a cryptography
_library_ then becomes a mechanical operation: implement the named
requirements in terms of already-existing, proven correct libraries.
Whether this becomes a boost library or not is not particularly
important, as most users do not and should not be casually reaching
for cryptographic primitives. For the limited set of users who do need
to use these primitives responsibly, accessing them from a library
that isn't Boost, is not a barrier.

Thanks