From: Daryle Walker (dwalker07_at_[hidden])
Date: 2003-04-25 01:19:00
On Wednesday, April 23, 2003, at 9:21 AM, Paul A. Bristow wrote:
> I feel Boost can also help by providing just one working
> implementation, even if just at 32-bit float accuracy, so any vendor
> who doesn't feel willing or able to provide a better one can still
> offer the Boost one and claim compliance.
If we start with just one version, wouldn't a "long double" version be
the best choice. That way, the "float" and "double" users can get
answers with quick & dirty conversions.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk