From: Stephen Nutt (snutt_at_[hidden])
Date: 2002-09-03 19:13:36


I took a brief look at thw fixed point work going on in J11/WG14, but
confess I've yet to read all 91 pages. As I understand it there will be two
new types, _Fract and _Accum both of which can we modified with
signed/unsigned and an integer size (short, long, int etc.) I believe the
class I propose will be functionally similar and will differ mainly in the
declaration of the variable.

Using a template will allow us to specify the underlying storage type of the
data, signed int, unsigned short etc. While it could be done with a union,
the code will become more complex, and there will be additional storage
beyond the required amount for smaller types (char, short.)

> This
> would seem to be a natural for templates. For example, since we
> couldn't implement the pragmas, could sat/modwrap and 1-or-2 ulp
> accuracy be policies?
Is a policy a template parameter?
I didn't know there was a cheap and easy way to get 2ulp. All my
calculations are currently 1ulp but I can reconsider my multiply and divide
for an optional and faster 2ulp.

Thanks for all the great feedback from all of you,

Steve