Re: unexpected rounding behaviour of mpf_add(). bug?

[Kevin Ryde]

Frederik Schaffalitzky <address@hidden> writes:
> 
> Yes, that would be more efficient. It means bumping the exponent down
> quite a bit, though.

True, but the spot of arithmetic for that has to be done anyway since
the operands might have a higher precision than the target.

> If I understand correctly, for a requested precision
> of 32*m bits, the number of 32-bit words allocated is m+2: one extra in
> case of carries and another extra one because the most significant word
> might be mostly zeros.

Yes, and _mp_prec is m+1, since the limb for carries doesn't count
towards intended precision.  I'll write a few words to clarify this
stuff, either in mpf/README or maybe a section on internals in the
manual.

> Is it true that mpfr will be fully integrated in future releases of GMP?

It'll be updated in the next release, but will remain separate.
Integration is the plan, once binary compatibility can be guaranteed.