Re: [boost] Math Constants Library formal review results

From: Ed Brey (brey_at_[hidden])
Date: 2001-04-25 12:19:17

From: "Peter Dimov" <pdimov_at_[hidden]>
>
> > I feel your pain; I'd like to have the simpler interface, too.  The
> > problem I see is that it is not possible for an implementer to
choose a
> > single best implementation for all users of the platform.  Some
users
> > will want constant folding and some will want binary (i.e.
hand-crafted)
> > precision.
>
> In the same application?

Good question.  I'm not really qualified to answer it, but if I would
pretend for a second that I am, I'd say that it doesn't seem too far
fetched to have an application that would generally want the
speed/convenience of constant folding, with a couple of algorithms here
and there that need maximum accuracy and are willing to pay the
speed/convenience price.

Pretenses aside, I'll happily defer to the wisdom of those with more
experience in this domain than me.