Re: [boost] Math Constants Library formal review results

From: Ed Brey (brey_at_[hidden])
Date: 2001-05-18 07:51:40

From: "Paul A. Bristow" <pbristow_at_[hidden]>

> I think people will expect to find names for pi/2 and 2*pi,
> and the programs will read more nicely if these are present.
> Of course a rational naming scheme is essential, as we have agreed.

My expectation is that code will be easiest to read if we avoid factors.
For one, I think seeing the math directly is more straightforward, as
in:
    pi / 2     and     2 * pi
vs
    half_pi     and     two_pi

The issue becomes more accute when the user needs constants that the
library doesn't provide.  He might end up with inconsistent code like
this:
    half_pi * sqrt( a * (pi / 5) )

If it were really all that notationally convenient to have
specially-named common factors of fundamental constants, I think you
would have seen the mathematics world a long time ago come up with
special symbols for 2 * pi, etc.

> Does anyone know what the costs are?
> Does it matter if we add a hundred names?
> Does it cause code bloat?
> Does it case linker constipation?

Under VC6, inline functions and namespace-level constants cost nothing
if you don't use them: they never even make it to the object file.
Static class members make it to the object file, but get stripped out by
the linker if they are unused.