Jeremie Pelletier:
>And I always thought it was double-extended 80-bit. The IEEE standard actually 
>specifies >= 79 bits and Intel implements it as 80bits.<

But it seems there are weird differences across operating systems regarding how 
much memory is allocated to represent such numbers, like 10, 12, 16 bytes on 
Windows, Linux, Mac (this comes from something Walter said when he has adapted 
DMD to the other OSs). Do you know why they have chosen different solutions 
regarding something so basic? (I don't even want to talk about the stupid 
endianess of CPUs).

Bye,
bearophile

Reply via email to