Re: [Algorithms] Complexity of new hardware

[Jon W. <jw...@gm...>]

Jason Hughes wrote:
> known offline.  Pawn also was a register machine rather than a stack 
> machine, so it could be compiled to native code and translated to very 
> fast code, compared to most other JITs. 

Is that really a difference? You can transform between the two systems, 
so they are semantically equivalent, although some constructs have more 
straightforward transforms than other. That, and the fact that the 
x86/32 is almost a stack machine already ;-)

>  LUA worries me a bit because 
> I've heard it's a little fast-and-loose with memory, though I imagine 
> that depends a lot on the way scripts are written.
>   

I've loved Lua to death for a long time for its smallish code size, its 
straightforward C integration, its clear separation of interepreter 
context and its lightweight approach to closures. However, despite that, 
I've never come to really like the Lua syntax. It's just slightly too 
weird to look good. Meanwhile, Python has always had a nice library and 
light-weight syntax (different enough to be unique, for sure!), but its 
C integration and size are quite cumbersome, and trying to do closures 
from C is like trying to pull teeth. (No, boost::python doesn't much 
help I'm afraid)

So, how to design a well-adopted, robust, small, powerful language with 
pretty syntax and an effective C integration layer? And what language 
has all that but I've missed? :-)

Sincerely,

jw