Programmable delimiters
Eduardo Cavazos
On the Ikarus mailing list, I brought up the subject of programmable
delimiters:
http://ikarus-scheme.org/pipermail/ikarus-users/2009-December/000692.html
Chibi, being so small and simple, was easy to coax into allowing the
'brackets' idea I mention there. :-)
>From an actual Chibi session:
> [list 1 2 3]
ERROR: undefined variable: brackets
;; Remember, [...] expands into (brackets ...). So, let's define
;; 'brackets' to be a macro such that [...] act like parenthesis:
> (define-syntax brackets
(syntax-rules ()
((brackets expr ...)
(expr ...))))
;; And try again:
> [list 1 2 3]
(1 2 3)
;; Now let's try a different 'brackets' macro which enables a shorthand
;; syntax for vector element getting and setting:
> (define-syntax brackets
(syntax-rules ()
((brackets v i)
(vector-ref v i))
((brackets v i val)
(begin
(vector-set! v i val)
v))))
> (define vec (vector 'a 'b 'c 'd 'e))
> [vec 0]
a
> [vec 0 'xyz]
#(xyz b c d e)
> vec
#(xyz b c d e)
>
I only needed to change 'sexp.c' and 'sexp.h'. I included the hacked
versions.
Ed
Alex Shinn
Eduardo Cavazos <wayo.c...@gmail.com> writes:
> On the Ikarus mailing list, I brought up the subject of programmable
> delimiters:
>
> http://ikarus-scheme.org/pipermail/ikarus-users/2009-December/000692.html
>
> Chibi, being so small and simple, was easy to coax into allowing the
> 'brackets' idea I mention there. :-)
Cute :) If you're going to use this it's probably worth
making {braces} programmable as well. And for real-world
use I think it may be important to ensure the closing
delimiters match their opening counterparts correctly, so
that [foo) would be an error.
... unless you wanted to make both names programmable so
that [0, 1) would read as (bracket paren 0 (unquote 1))
which could expand into a half-open set representation :)
--
Alex