An Object model that can be customised
to support powerful features such as multiple inheritance and
mixed-mode execution (native code/ byte code)
I also guess that it may help to implement dynamic class loading;

The new D object model ?
Bjoern

You will find a 144 LOC !  C implementation at the end of this document:
http://www.piumarta.com/pepsi/objmodel.pdf

and an implementation in Python at :
http://denkspuren.blogspot.com/2007_03_01_archive.html

BLS Wrote:

> An Object model that can be customized
> to support powerful features such as multiple inheritance and
> mixed-mode execution (native code/ byte code).

Really working on those complete sentences there?

But, sure, it's not a bad idea (though, IMO, multiple inheritance _is_). But it seems like it'd be a lot of work at this stage for Walter, and I certainly don't think it should become part of the spec.

Robert Fraser schrieb:
> BLS Wrote:
>
>> An Object model that can be customized
>> to support powerful features such as multiple inheritance and
>> mixed-mode execution (native code/ byte code).
>
> Really working on those complete sentences there?

Yes, A smalltalk derivat is allready implemented based on this object model, which is, compared to Cpp, approximate /only/ 20 percent slower.

>
> But, sure, it's not a bad idea (though, IMO, multiple inheritance _is_). But it seems like it'd be a lot of work at this stage for Walter, and I certainly don't think it should become part of the spec.

Instead of multiple inheritance implementing *traits* is also possible :

quote :
Traits are a powerful software composition mechanism. A trait
is a collection of methods without state that can be manipulated and
combined with other traits according to an algebra of composition,
aliasing and exclusion. They are interesting because they provide
the power of multiple inheritance without the drawbacks and complexity
of multi inheritance ....
end quote


Bjoern

BLS schrieb:
> Robert Fraser schrieb:
>  > BLS Wrote:
>  >
>  >> An Object model that can be customized
>  >> to support powerful features such as multiple inheritance and
>  >> mixed-mode execution (native code/ byte code).
>  >
>  > Really working on those complete sentences there?
> 
> Yes, A smalltalk derivat is allready implemented based on this object model, which is, compared to Cpp, approximate /only/ 20 percent slower.
> 
>  >
>  > But, sure, it's not a bad idea (though, IMO, multiple inheritance _is_). But it seems like it'd be a lot of work at this stage for Walter, and I certainly don't think it should become part of the spec.
> 
> Instead of multiple inheritance implementing *traits* is also possible :
> 
> quote :
> Traits are a powerful software composition mechanism. A trait
> is a collection of methods without state that can be manipulated and
> combined with other traits according to an algebra of composition,
> aliasing and exclusion. They are interesting because they provide
> the power of multiple inheritance without the drawbacks and complexity
> of multi inheritance ....
> end quote
> 
> 
> Bjoern

For a better understanding have a look at http://vpri.org/pdf/colas_wp_RN-2006-001-a.pdf

See graphic at page 5 and than  page 23

More at :
http://vpri.org/html/writings.htm
Chapter *Fundamental New Computer Technologies*

Okay that is more than enough <g>