long doubles

[Damien Stehle]

Hello,

I wonder why GMP it is not possible to perform conversions between
mpz_ts and long doubles. More precisely, why are the following
functions unavailable?
mpz_get_ld 
mpz_set_ld
Their task (and possibly their code) would be to copy those of
mpz_set_d and mpz_get_d. 

Besides, why does mpz_set_d rounds towards zero and not to the nearest
(with the round to even rule)?

Regards,
Damien