Re: Integers and other numbers [was Re: dlw_int review]

From: Daryle Walker (darylew_at_[hidden])
Date: 2001-09-27 18:41:47

on 9/27/01 6:51 PM, helmut.zeisel_at_[hidden] wrote:

> --- In boost_at_y..., "Eric Ford" <eford_at_m...> wrote:
>>> Other types like dimensioned quantities, std::complex, quaternions,
>>> polynomials etc. are more or less independent from that roadmap and might
>>> probably have a template parameter for the respective numeric type.
>> 
>> The important thing is that the various types need to be able to work with
>> each other.  Things like a matrix of complex numbers with dimensions of time,
>> units of seconds, implemented in terms of a fixed precission floating point
>> type need to work.
> 
> Yes.
> 
> But this is not specific to numeric types but true for the whole library.
> 
> E.g., for me it is quite a lot of work to make the formating of bigint
> consistent to iostream.

I realized that all the integer classes we make would have this problem, so
I made a general routine that all of us can use.  Look at "int_read.zip" in
the vault.

> OTOH, the interface boost::rational requires from bigint is (with only one
> exception: rational_cast) naturally fulfiled from what I considered to be the
> correct interface for big int.
> 
> Anyway, when we proceed on the road, it will be necessary from time to time to
> adapt the interface of one class to satisfy the needs of some other class.

-- 
Daryle Walker
Mac, Internet, and Video Game Junkie
darylew AT mac DOT com