On 02/01/15 10:59, Manu via Digitalmars-d wrote:
Don't know if it's entirely the same issue, but Don C. and I have been
having occasional (admittedly fairly casual) chats about the need to do
something about std.math and its use of real. I don't want to make any
strong promises (far too much going on work-wise), but I could well be
persuaded to look into that with some seriousness, if it could be helpfully
related to std.simd work.
I'm sure it would.
What do you mean 'do something about real'? You mean that x87 is
deprecated for more than 10 years now, and will probably go away at
some point.
I mean the fact that std.math functions on floating point are hardcoded to use
real for their input and return types.
