Re: [Bug-gsl] enforcing the order of numerical operations

[Brian Gough]

At Wed, 25 Jul 2007 22:59:14 -0700,
Eugene Loh wrote:
> Anyhow, I'm looking for guidance as to what you'd support in
> putbacks.  Do you favor the "volatile" approach?  Are you open to
> the dummy() function?

I think the best approach would be to use a macro, so the definition
could be changed as necessary -- this is what we do for avoiding
excess precision on x86, see the definition of GSL_COERCE_DBL(x) in
acconfig.h for example.  It would make sense to have a different macro
for forcing an evaluation, maybe GSL_PROTECT_DBL(x) or something like
that.  Previously I didn't do anything systematic as I only
encountered a few problems where volatile was needed.

-- 
Brian Gough