Re: What's right with FORTH and wrong with c.l.f.

mhx@[EMAIL PROTECTED]
 (Marcel Hendrix) wrote Re: What's right with FORTH and wrong
with c.l.f.
[..]
>                                      E.g. if the input is 

>  12 apestaart 3456 kippekot 18 hondehok -771

> the numbers can be fetched by string parsing and >NUMBER.

> But also with  

>  : apestaart ; : kippekot ; : hondehok ; PARSE-NAME EVALUATE

Free associations .. Would it be nice to have a special type of VOCABULARY
which does *not* insist on a direct string match, but accepts
patterns (vectored FIND)? We could then define not only
words with names, but also words with names that match a certain
pattern. Let's say we can do that with "::" instead of ":".

VOCABULARY myapp  myapp DEFINITIONS 
  S" [a-zA-Z]+" :: ( noop ) ;
PREVIOUS DEFINITIONS
( you must be careful not to leave myapp in the normal search order :-)

12 apestaart 3456 kippekot 18 hondehok -771

would now be equivalent to 

12 3456 18 -771

Or 

VOCABULARY myapp myapp DEFINITIONS 
  S" ," :: ( noop ) ;
  S~ "[a-zA-Z]+"~ :: last-name 1- 1/STRING 2DUP $CREATE ( magic1 ) , (
magic2 ) , ;
PREVIOUS DEFINITIONS

... where last-name ( -- c-addr u ) returns the matched pattern. This
allows

"HARRY","MAGDA"

... to become Forth words. Magic1 adds strings to a sorted list, 
returning the new head, and magic2 returns n based on some formula. 

Must try this in my copious spare-time.

-marcel