mpf_get_d

[Paul Zimmermann]

> This was fixed a while ago:
> 
> http://shell.gmplib.org:8000/gmp-4.3/rev/32cc41c2fdea
>   
> Please try the patch.

thanks, this indeed fixes the problem (I had checked on gmplib.org for
"Current release status" but no issue was reported for 4.3.1, should we
look at another place for known issues?)

Paul

PS: the documentation of mpf_get_d says that for "too big" input an infinity
is returned. It would be more coherent with the "truncating" mode to return
the largest representable floating-point number (or its opposite).