
Boost : 
From: Paul Mensonides (pmenso57_at_[hidden])
Date: 20030421 16:46:16
Vesa Karvonen wrote:
> I'm also very interested in seeing the Chaos library from Paul. I do
> not really know the philosophy of Chaos to be sure, but I think that
> Order and Chaos could benefit from and complement each other. The
> Orderinterpreter
> has been made possible by several discoveries in PPmetaprogramming
> by Paul and me.
Chaos *is* the pplib, but redesigned from the ground up to take advantage of
strictly conformant preprocessors. It also includes facilities that are
significantly more complex than the CVS pplib. Highprecision arithmetic is
implemented for example, as is a parametric lambda facility (not a
lambdacalculus facility like Vesa's work, more like the other Boost lambda
libraries). The most fundamental differences between Chaos and the CVS pplib
is a full abstraction of recursion. It is *algorithmic*, which means
implementing things like REPEAT, FOR, FOLD_LEFT, WHILE, and even
FOR_EACH_PRODUCT all only require about three or four macros each. Chaos is
also implemented from the ground up to support variadic macros and placemarkers
from C99.
> Also, Paul has mentioned that Chaos implements efficient high
> precision arithmetic. The high precision arithmetic of Chaos could
> probably be used from the Orderinterpreter, which could, for some
> parts, basically act as a convenient highlevel frontend to the
> efficient lowlevel facilities of Chaos. But like I said, I don't
> know enough about Chaos to be sure.
I'm still debating whether to provide fixed, highprecision arithmetic, or
arbitrary precision arithmetic. I can do either, but fixed orecision would be
faster. Maybe I'll do both....
Regards,
Paul Mensonides
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk