Raul wrote:
> Ok: it is precisely when you change problem domains where
> one mechanism becomes superior to another.
Precisely. In fact, if you read the thread I cited in my previous message,
you'll see that I entirely agree with you when it comes to J scripts.
I state that J "workspaces" (aka locales) are better stored in script files
(aka J code aka 5!:5 ) than in binary format (aka 3!:1 ).
My point was only that for complex J structures (large, deeply-nested J nouns),
the 5!:5 representation is not really more friendly than the 3!:1
representation.
w =: {.@:;:
cjs =: 5!:1 w 'toJ' NB. Complex J structure
5!:5 w 'cjs'
<(<'@:'),<(<(<,'3'),<(<(,'0');10{a.),(<(<,'}'),<,<(<,'@'),<(<(<,'@'),<(<'I.'),<(<,'&'),<(<'e.'),<(,'0');13{a.),<,']'),<,']'),<(<,'2'),<(<(<,'~'),<,<,'#'),<(<,'@'),<(<'-.'),<(<,'@'),<(<(<,'&'),<(<<;.(_1)
32 48 32 13 10{a.),<'E.'),<,','
> I was attempting to discuss the idea that the one mechanism
> might be universally better than the other.
"Universal" is a broad term. If arbitrary domain changes are permitted, it is
possible that neither 5!:5 nor 3!:1 is superior, or even permissible.
In fact, if arbitrary domain changes are permissible, one might question
whether discussion is fruitful.
-Dan
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
