Re: [boost] Crypto Proposal

Subject: Re: [boost] Crypto Proposal
From: Scott McMurray (me22.ca+boost_at_[hidden])
Date: 2010-04-14 13:02:56

On 14 April 2010 12:44, Chad Seibert <chadseibert_at_[hidden]> wrote:
>
> Writing this as SP would look like the following (if I understand SP
> correctly):
>

For static polymorphism, you write to provide the interface demanded
by the concept; you don't need to derive from anything.

Consider the Boost.Random concepts, for instance
<http://www.boost.org/doc/libs/1_42_0/libs/random/random-concepts.html>.
 boost::mt19937 isn't derived from some UniformRandomGenerator class.
It just provides the necessarily facilities.

In my experiments with hash functions, the nicest parts of generic
programming has come from the various constructions used to build
algorithms out of primitives.

For example, I can write a SHA-512 block hash as

    merkle_damgard_block_hash<
        sha2_policy<512>::initialization_vector_generator,
        davies_meyer_compressor<shacal2_block_cypher<512>, state_adder>,
        digest_from_state<digest<512>, display_endian::big_word_big_byte>
    >

By building it out of the SHACAL-2 block cypher
<http://en.wikipedia.org/wiki/SHACAL> run through the Davies-Meyer
construction to produce a One-way Compression Function
<http://en.wikipedia.org/wiki/One-way_compression_function#Davies-Meyer>
which is used repeatedly in the Merkle–Damgård construction to create
the hash function
<http://en.wikipedia.org/wiki/Merkle%E2%80%93Damg%C3%A5rd_construction>.

There's no BlockCypher class, no OneWayCompressor class, etc.  They're
just concepts.