Subject: Re: [boost] Fixed point integer proposal
From: Kenny Riddile (kfriddile_at_[hidden])
Date: 2009-06-25 16:26:38
Daniel Hulme wrote:
>>> My use-case for fixed-point arithmetic is game development on embedded
>>> devices that lack FPUs. Native support for trigonometric functions is
>>> something I would definitely want.
> On Thu, Jun 25, 2009 at 01:57:51PM -0500, Soren Holstebroe wrote:
>> I can see there could be a need for it. However, in my opinion, it
>> would be out of the scope for a fixed point template to accurately and
>> effectively implement trigonometric functions.
> Regardless, it's easily something that can be added later, and whenever
> it's added, it makes sense to put it in a separate header from the class
> template anyway. I'm sure that at review time, the people who want a
> useful system now that can be built upon later with extra functions will
> outvote anyone who would rather hold their breath for something that
> does everything everyone might want.
Agreed. I was just saying that there are use-cases where simply
converting to float and back isn't good enough.
Boost list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk