Though I initially mis-interpreted this thread as an argument for a
throwback
to order-of-operation parsing, now that I better understand it, I remain
unconvinced
this proposal's usefulness outweighs its complication.

Consider that the original example would be better written in a more J-like
fashion, without parens, like this:

   1-~4 2+/ . * 3 3

(assuming the two "3"s only accidently have the same value and avoiding the
special decrement function "<:" on the same assumption for "1".)

It's hard to abstract from one simple, scalar, example like
this to understand the advantage of changing the parsing order,
especially one which places such heavy syntactic lifting on
an invisible character.

This reminds me of a lecture on C++ by Bjarne Stroustrup I attended a few
years ago.
I ran into a friend from the APL community there and we sat next to each
other.  Bjarne
began talking about the advantages of overloading certain symbols.  When he
started
showing examples where he was overloading the " " (space) operator, my
friend and
I both looked at each other in amusement as we simultaneously realized that
the
date was April 1st and Bjarne was having us on.

But it's a little too early for that to be the case here.

Devon

On 3/23/06, p j <[EMAIL PROTECTED]> wrote:
>
> I assume that was directed to me.
>
> Many of you will somehow ignore the next sentence, or
> disbelieve my sincerity regarding it.  I fully
> understand the reason and practicality for right to
> left parsing in J, and have no desire to change it.
>
> Having the ability to override the default parsing
> order is an undisputed good thing.  Working with
> parens is usually a needlessly complicated way to
> override that parsing order.  The proposal is to allow
> the programmer to use a single token ' ' in situations
> that currently require 2 matching '()'.  Its hoped
> that the option makes code clearer and easier to read
> write and edit.
>
...
--
Devon McCormick
^me^ at acm.
org is my
preferred e-mail
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to