On 11 December 2014 at 16:41, Maciej W. Rozycki <ma...@codesourcery.com> wrote: > -- notice the reintroduction of the different, wrong default NaN encoding > patterns. Will you be sending an update shortly so that I can schedule > my MIPS IEEE 754-2008 testing tonight?
Ah, yes, as you point out these are all actually not NaN bit patterns at all. I've just sent out a fixup patch which makes them all consistently sign-bit-clear, exponent-bits-set, top-mantissa-bit-clear, other-mantissa-bits-set. (I'll squash that patch into the other one for v2.) thanks -- PMM
