RE: [boost] Math constants for naive and gurus? - which constants do you want?

From: Fernando Cacciola (fcacciola_at_[hidden])
Date: 2001-10-24 15:46:06

----- Original Message -----
From: Peter Dimov <pdimov_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Wednesday, October 24, 2001 4:24 PM
Subject: Re: [boost] Math constants for naive and gurus? - which constants
do you want?

> From: "Fernando Cacciola" <fcacciola_at_[hidden]>
> [...]
> > The user of my kernel libraries might want to use his ReallyBigFloat
type,
> > and he might want my kernel to use the ReallyBigFloat_PI that he has, so
I
> > need a mechanism to allow him to specify the values of the generic
> constants
> > up to the full precision of the generic type.
>
> I agree; it's only that I'm suggesting that, perhaps, allowing such user
> control is your responsibility, not boost's; partly because there probably
> are more aspects of user customizability than the value of 'pi'.
>
I see.

I know that currently, boost has not already undertaken the issue of generic
math programming systematically.
Therefore, in the context of the current boost, I agree that fixed type
constants are better fit into boost than generic template-based constants.

However, I would like too see boost starting to take generic math
programming into account as soon as possible.
For example, Eric Ford, Kevin Lynch and I are working on a generic version
of the standard math library; thus, if that work makes into boost, the
template based generic constants would fit very well.

IMO, this issue can be layered, having a fixed type layer (with suffixes for
the float and long double cases) using a straight constant; and a generic
layer, using some of the template based version suggested. In fact, the
generic layer could take the values from the fixed-type layer.
Now, if those going to actually write this stuff want to provide both
layers, great!, but I can live with only the fixed-type layer in the worst
case.

Fernando Cacciola
Sierra s.r.l.
fcacciola_at_[hidden]
www.gosierra.com