On Mon, Jul 13, 2015 at 7:50 PM, David Jeske <[email protected]> wrote:
>
> On Jul 13, 2015 8:26 AM, "Keean Schupke" <[email protected]> wrote:
>> I agree until the "unlike C++". Clearly you can implement all this in C++,
>
> This argument is not helpful to understanding c++ capabilities.
>
> Turing completeness already says we can implement anything in any of these
> systems, even if it means implementing the other system. We are talking
> about how different languages solve problems in their native constructs, not
> what one can theoretically implement (which is the same for all turing
> complete languages).
>
> What we are talking about is how the native language structures are handled
> by the runtime.
>
> In known C++ compilers, base class size, vtable layout, parametric
> instantiation and instance variable layout are hard coded at compile time.
> In objc/swift/clr/java, these are all resolved dynamically at load time.
> This enables many binary compatible dll updates which are not possible in
> c++ (* when actually using c++ language features)

Just to add to what David is saying, If the native language does not support it,
the simplest way to implement it is going to be through unsafe pointer
arithmetic
and implementing it in a safe language which doesn't support it is
likely to end up with something completely alien to the host
language...
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to