Re: Targeting the Lua VM

[therandthem <therandthem@...>]

I can't speak to the abilities of LuaJIT.  Yes, it is
the fastest thing around (for dynamically typed
languages) and you could certainly do worse by
targeting your own VM.  Several languages have done
worse (Python, Ruby, JavaScript).  Some VMs out there
are so bad that I cannot imagine LuaJIT being worse. 
I don't know why 80-times slower than C is acceptable
to some people.

That being said, I would suggest you see what you can
accomplish with meta-programming in Lua as a language.
 Maybe you can get all the features you are looking
for and most of the syntax without having to get into
the lower levels of LuaJIT.  Let me give you a few
resources to get started.

http://livelogix.net/logix/  Live Logix was some
Python software that set out to create a
meta-programming environment where you could write new
languages, DSLs and recreate existing languages all
through meta-programming.  It seems to have failed,
but it is still a great idea.

http://angg.twu.net/miniforth-article.html
Boostrapping a Forth in 40 lines of Lua code.  Nice
article on using Lua rather than C for creating your
next language.  It argues that Lua (as a VM and a
language) is way more accessible than C.  It shows you
some Lua meta-programming techniques.

Cheers,

McKinley


--- KHMan <keinhong@gmail.com> wrote:

> I think CLR and Parrot already solves the multiple
> scripting
> language problem in a more practical way. After
> fixing up the Lua
> VM to implement various languages efficiently, you'd
> probably end
> up with something like CLR or Parrot. So, why
> reinvent the wheel?
> Something with Lua VM is probably better for
> language subsets, so
> for general interpretation of language sources, it
> won't fit the bill.


       
____________________________________________________________________________________
Yahoo! oneSearch: Finally, mobile search 
that gives answers, not web links. 
http://mobile.yahoo.com/mobileweb/onesearch?refer=1ONXIC