Re: should GOOPS be built-in?

[Valentin Kamyshenko <val at kamysh dot materials dot kiev dot ua>]

>>>>> "Mikael" == Mikael Djurfeldt <mdj@mdj.nada.kth.se> writes:

Mikael> E.g., if one realizes that the arithmetic operators in R5RS
Mikael> can be regarded as generic functions, representing them as
Mikael> such internally will make it possible to 1. give a cleaner and
Mikael> more modular structure to the numerics code, with less
Mikael> interdependencies, and 2. provide a natural way to "plug in"
Mikael> new functionality such as support for GMP bignums and/or
Mikael> rationals.
For sure, this will simplify things a lot. I tried few times to go
through the numbers.c (thinking about adding <rational> to it) , and 
I must admit that I'm too lazy to change there anything :). But, when 
regarding such a fundamental feature as the numbers tower, will not
using GOOPS at this level lead to loss of performance? I know, this
question was answered few times, but I was not convinced, because
comparison seemed to be "at the scheme level", not "at the C
level". Does not it worth to experiment with GOOPS-based number tower
before making such a fundamental decision? By the way, concerning the
documentation:
	1. Are there available any examples of GOOPS programming "at
the C level"?
	2. Where can I find a description of MOP (there is no
reference in the GOOPS manual)? 

--
Best regards,
	Valentin.