mpz_get_d() with rounding

[Jeroen Demeyer]

On 2014-05-22 11:30, Torbjörn Granlund wrote:
> ALso, I expect this to be tricky to implement in a portable and correct
> way.
Are you willing to make the assumption that doubles have 53 bits of 
precision? If you do, it's easy enough to implement.