Re: [hash2] Difference from paper: HashAlgorithm interface vs. function object

From: Peter Dimov (pdimov_at_[hidden])
Date: 2024-12-12 14:57:08

Joaquín M López Muñoz wrote:
> > No dia 12 de dez. de 2024, Ã s 14:22, Peter Dimov via Boost
> <boost_at_[hidden]> escreveu:
> >
> > Using default-constructed instances of the hash algorithm, instead of
> > copies, makes the hashing seed-independent, which means that a way to
> > engineer collisions will work regardless of the seed used.
> 
> Can’t you construct h2 by seeding it with a result from h (possibly after
> updating h with the range size)?

That's (a) much slower and (b) requires the ability to obtain a result without
perturbing the state, which is basically equivalent to making a copy.