Re: [colorforth] DOES> How is colorForth different from other Forths?

[Jonah Thomas <j2thomas@xxxxxxxxxx>]

Frédéric DUBOIS wrote:
> De : Chuck Moore [mailto:chipchuck@xxxxxxxxxxxxxx]

> > Trade-offs were different with threaded code. But when 
> > compiling native
> > code, fewer syntatical constructs seems better. It's nice to say that

> Why?

I can't speak for Chuck, but I can say something.  With threaded code 
you get a real hardware advantage from DOES> .  It does some magic that 
you can't do with straight Forth code.  If you try to write DOES> in 
high-level Forth code that doesn't use DOES> you get something obviously 
inefficient and clunky.  If it's clunky and inefficient with native 
code, why do it?

> > colorForth source has a 1-1 correspondance with object code. 
> > If several
> > syntaxes generated the same code, it would be a many-1 correspondance.

> How much do you value this correspondance? Is there some benefits to use it
> in bytecode implementations for example?

I'm not clear what Chuck will use in place of DOES> .  There might be 
something worth doing.  When you make a macro it does a series of 
instructions from one name.  Traditional DOES> is more complicated than 
that, I don't think you can write it so it just compiles a short series 
of instructions.  But you might make macros that do <literal> <command>. 
 That's the final result that DOES> gives you, except that the literal 
is something special, a particular address.  It doesn't have to be an 
offset from HERE , it could be anything that's useful to you, the 
address was just the only way that DOES> did it.  So if you do that 
particular style of macro often enough it might be worth making a macro 
to make that kind of macro.  You might have something like

( x ) ' <command> DOIT <name>

I don't know whether that's worth doing.  I sure don't know whether it's 
worth doing it more like DOES>.

' <command> MAKE-DO MY-PARENT
( x ) MY-PARENT MY-CHILD

Now you have a hierarchy of macros, the MAKE-DO word, a different 
MY-PARENT variation for each command, and then a different child for 
each address etc.  Do you need that many words?  A little more typing 
gives you every final result with just the one DOIT word plus a name for 
each child.   And for just a : and a ; and a couple of color changes you 
get that final result with no scaffolding at all.

You'll have to find out by experience how many extra words are worth having.

Best wishes.



---------------------------------------------------------------------
To unsubscribe, e-mail: colorforth-unsubscribe@xxxxxxxxxxxxxxxxxx
For additional commands, e-mail: colorforth-help@xxxxxxxxxxxxxxxxxx
Main web page - http://www.colorforth.com