I wrote:
>  I found an example where this specification was not honored, and 
>  wanted to know why, or what I was missing when reading the DoJ. [1]

Raul responded:
>  I am not aware of this example. 

The example I provided in my original post demonstrates the discrepancy only 
under my assumptions.  In particular, my code
demonstrates the discrepancy only assuming that "proverb" is defined as "a 
[mutable, non-primitive] name that refers to a verb",
and specifically doesn't demonstate it using your proposal that "proverb" is 
defined as "a [mutable, non-primitive] name that
doesn't refer to a noun" (which isn't surprising, given that you defined it 
that way to avoid the discrepancy).

Let me put it more concisely.  If "proverb" meant exactly "a [mutable, 
non-primitive] name that refers to a verb"  [1] and only
that, would you agree there would be a discrepancy between the specification 
(J) and the implementation (j)?

If not, then please describe why.  That is precisely the information I am 
seeking (i.e., the reason I asked the question in the
first place).  If so, then I need to do that research & cogitation I mentioned 
to determine if mine is the (only reasonable)
definition of "proverb" (given the text of the DoJ and no supplementary 
material). [2]  

I wrote:
> Victor and Raul's responses corroborated my own observation 
> that all names, regardless of nameclass, are resolved when executed, but not 
> before (in particular, not as they're stacked).

Raul responded:
> Except, of course, that nouns are executed when they are stacked.
> They need no parameters -- they just need a context where 
> their value is relevant.

For the sake of rule minimization (consistent brevity), I am treating nouns 
like other nameclasses -- executed when provided the
right number of arguments -- in this case, zero.  I kind of like the 
nameclasses hierachy this introduces.  Perhaps this is
equivalent to having an "implicit" production rule that  N  stacks  N  ?  
Anyway, this detail isn't relevant to the current
discussion.

-Dan

[1] The "[mutable, non-primitive]" part is an attempt to restrict our 
discussion to user-defined names -- it may not be precise.
Let's not quibble about that part (though it would be an interesting discussion 
in itself).

[2] If do arrive at that conclusion, I will post my reasoning for the 
consideration of the Forum and the maintainer of the DoJ.

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to