Re: Problems with square brackets

[Jamison Hope <jrh at theptrgroup dot com>]

On Aug 31, 2011, at 8:02 PM, John-Paul Verkamp wrote:

> In 'CHANGES FROM KAWA 1.6.56 TO 1.6.58' it's mentioned that 'Square
> brackets [ ... ] are allowed as a synonym of parentheses ( ... ).'.
> But when I try Kawa (either using eval(...) or using the Kawa REPL
> directly, I get this:
>
> #|kawa:1|# (let ([x 5]) x)
> /dev/stdin:2:1: let binding is not a pair:[x
>
> The same thing happens for cond, etc.
>
> Is there a way to enable this?
>
> I'm particularly interested in it working correctly when using (load
> "..."). When I'm using eval(...) I can just do a quick replace to work
> around the issue.
>
> Thanks!

A while back, (after the release of 1.11, I think), Per changed the  
default
bracket behavior to facilitate the construction of immutable literal  
vectors
or (Java) arrays. You can change this by calling  
ReadTable#setBracketMode(),
which says:

  /** Specify how '[' and ']' (and '<') are handled.
   * The value -2 means {@code [a b c]} is {@code ($bracket-list$ a b  
c)}
   * and {@code f[a b]} is {@code ($bracket-apply$ f a b)}.
   * The value -1 means that '[' and ']' are plain token constituents.
   * The value 0 means that '[' and ']' are equivalent to '(' and ')'.
   * The value 1 means that '[' and ']' are equivalent to '(' and  
')', except
   * within a token starting with '<', in which case they are  
constituents.
   * This is so '[' is non-terminating when reading say '<char[]>'
   */

The default value is currently -2. If you change it to 1, then Racket- 
style
brackets==parens will work (though I think Racket limits the meaning of
brackets to certain contexts, like within a let; this will make them
equivalent everywhere):

#|kawa:1|# [display "hello"]
#(#<procedure display> hello)
#|kawa:2|# (set! (gnu.kawa.lispexpr.ReadTable:get-current):bracket- 
mode 1)
#|kawa:3|# [display "hello"]
hello
#|kawa:4|# (let ([a 3] [b 4]) (sqrt (+ (* a a) (* b b))))
5.0

I think if you limit yourself to more-or-less portable Scheme code, you
should be fine with that, it'll just cost you some brevity if you do
Kawa-specific coding where $bracket-list$ is a handy shortcut.
For instance, instead of being able to write
(define my-array-of-doubles ::double[] [1 2 3 4 5])
you'll have to write
(define my-array-of-doubles ::double[] (double[] 1 2 3 4 5)).


Per, setting the bracket mode seems like a plausible candidate for a  
"#!"
special form a la #!fold-case.


-Jamie

--
Jamison Hope
The PTR Group
www.theptrgroup.com