On 9 Jan 2009, at 09:50, Yen-Ju Chen wrote:

> Floating value doesn't work for
> me.

This should probably be documented more clearly.  Supporting unboxed  
floats would be messy in Smalltalk, and I don't think it's a high  
priority since you probably don't want to be using Smalltalk for  
computationally expensive tasks.  What I would like to do is create  
two BigInt subclass, which wrap the GMP arbitrary precision rational  
and floating point stuff, and have BigInt and its subclasses  
automatically create the correct kind for the results of calculations,  
so you only ever lose precision of you do something (e.g. sqrt) which  
results in an irrational number.

David

_______________________________________________
Etoile-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/etoile-discuss

Répondre à