Henry;
Thinking that Jsoftware's products will not undergo seemingly capricious
changes over time shows one's total lack of understanding of how Jsoftware
works. From the start, upward compatibility was a non-goal, in deference to
doing the right thing, a concept more aesthetic than rigorous.
------------------------------------------------------------------------
|\/| Randy A MacDonald | APL: If you can say it, it's done.. (ram)
|/\| [EMAIL PROTECTED] |
|\ | |If you cannot describe what you are doing
BSc(Math) UNBF'83 |as a process, you don't know what you're doing.
Sapere Aude | - W. E. Deming
Natural Born APL'er | Demo website: http://156.34.75.173/
-----------------------------------------------------(INTP)----{ gnat }-
----- Original Message -----
From: "Henry Rich" <[EMAIL PROTECTED]>
To: "'Beta forum'" <[email protected]>
Sent: Thursday, July 06, 2006 12:24 AM
Subject: RE: [Jbeta] Incorrect result shape from 5 +"1 (0 1 $' ')
I want my program to depend on J's following the rules.
Most of the problems I have had come from boxed nulls rather
than character, but the point is the same: if you don't follow
the rules, how am I supposed to write working code? Do I have
to check the result of every expression to see if you have
decided to add 'knowledge' to the interpreter that changes
a result?
You seem to be arguing that if an expression is unusual
enough I should be willing to accept whatever result you
find convenient. It's as if you told me that
3.3428953286 + 2.645451867 was implemented to be 6.0 .
It probably wouldn't hurt me, but I would think it
arbitrary and unjustified.
Changing the interpreter to follow the rules is fine; changing
it to deviate from the rules is abominable.
I don't think anything in the interpreter now causes spurious
trading. I hope that condition persists. And if not - well,
maybe the stock will go up.
Henry Rich
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of R&S HUI
Sent: Wednesday, July 05, 2006 11:06 PM
To: Beta forum
Subject: Re: [Jbeta] Incorrect result shape from 5 +"1 (0 1 $' ')
What you are arguing for is that
$ 5 +"1 (0 1$' ') should be 0 but that
$ 5 +"1 (0 1$0 ) should be 0 1 .
Is that really what you want your stock
trading program to depend on?
----- Original Message -----
From: Henry Rich <[EMAIL PROTECTED]>
Date: Wednesday, July 5, 2006 7:55 pm
Subject: RE: [Jbeta] Incorrect result shape from 5 +"1 (0 1 $' ')
> What I meant was, if I have defined
>
> plus =: +
>
> then +"1 and plus"1 should always get the same result.
> (That's why I wrote 'plus' rather than 'f').
>
> The interpreter doesn't need to know that + and plus are
> the same. All it needs to do is follow the rule for
> empties, which requires checking the type, because the
> cell of fills is typed.
>
> I don't remember how I got into looking at 5+"1 (0 1$' ')
> but that isn't important. The practical difference to
> me between 0 1$0 and 0$0 is that
> (0 1$0) , 0$0
> 0
> (0 1$0) , 0 1$0
>
> in other words, the wrong shape produces a list with
> one item when it should have 0 items.
>
> Maybe that won't matter - it might just be something
> on the screen a few seconds. Or, it might be a list
> of stock transactions to perform, in which case I will
> incorrectly commit 25% of my net worth to a bad stock.
> (This has actually happened. Last time the stock went
> up).
>
> So I really want rules to go by. Learning by experiment
> can be awfully expensive.
----------------------------------------------------------------------
For information about J forums see
http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
