Organization of multiplication functions

[Kevin Ryde]

Mark Sofroniou <marks at wolfram.com> writes:
>
> Fair enough. Are there any plans to officially document mpn_sqr_n,
> (and have it callable e.g. through a dynamic library)

Hmm.  Probably not.

> or is it considered
> an internal library function, so the preferred interface is intended to stay
> as mpn_mul?

mpn_mul is probably best for the higher level functions, since mpz_mul
etc can just call it and let it work out if we're squaring or
multiplying.

mpn_sqr_n probably only finds direct use in the various bits of
powering stuff.  If they're slow anyway then maybe it doesn't need to
exist at all.