gmp 4.2.4 - mpf_add two large negative floats - strange result

[Sisyphus]

----- Original Message ----- 
From: "Vincent Lefevre" <vincent at vinc17.org>

[snip]

>
> In GMP, the precision is a multiple of the size of a limb in bits
> (so, in practice, a multiple of 32 or 64). The default precision
> is __GMPF_BITS_TO_PREC(53), where:
>

Another "Ooops!!" on my part.
Setting the default precision to 104 in fact sets the precision to 128 (not 
104 as I had assumed).
For some reason I was thinking GMP set the precision to a multiple of 8, not 
32. (One of these days it might occur to me that I ought to take the extra 
few seconds, and actually check on the facts before posting.)

Cheers,
Rob