I wasn't clear.  I don't want to change the dictionary.  I want to suggest a 
use for the word "type" in the context of J.  "Leave it at that" was intended 
to mean, do not give "type" a broader meaning in the context of J.

Sent from my iPad


On Apr 23, 2012, at 12:15 PM, Don Guinn <[email protected]> wrote:

> Each atom of a boxed array can be any of the four types of a noun. I think
> the wording is to cover that case.
> On Apr 23, 2012 11:03 AM, "km" <[email protected]> wrote:
> 
>> From the Dictionary II.A. Nouns, "The atoms of any array must belong to a
>> single class: numeric, literal, symbol, or boxed."
>> 
>> How about saying the type of a noun is numeric, literal, symbol, or boxed.
>> Leave it at that.
>> 
>> Sent from my iPad
>> 
>> 
>> On Apr 23, 2012, at 8:24 AM, Raul Miller <[email protected]> wrote:
>> 
>>> "Type" has been an ill-defined term in the context of J.
>>> 
>>> But it's a popular word, for describing some aspects of computer
>>> programs, and a short word.  So it probably deserves some use.
>>> 
>>> And, I think, in the context of J, it should signify a thing that can
>>> go inside parenthesis.  That concept needs a name, and I think "type"
>>> fits it rather well.
>>> 
>>> Thus, we have system types: (1)(+)(1)
>>> 
>>> We have named types:  (smoutput)(ARGV)
>>> 
>>> We have anonymous types:  (+/%#) (i.9)
>>> 
>>> We have types of nouns, types of verbs and so on.
>>> 
>>> We have numeric types, textual types, list types (and these can
>>> overlap), and so on.
>>> 
>>> etc.
>>> 
>>> Now, one thing that this loses is the idea of a type as a
>>> collection... or does it?  (1 2 3 4)
>>> 
>>> No, what this loses is the idea of a type as a representation of
>>> possibilities... or does it?  ? bind 100
>>> 
>>> No, what this loses is the idea of a type as a storage format.  But
>>> when I have proposed the idea that "type" represents a storage format,
>>> I have been told that that's silly, by people who would be in a
>>> position to know.
>>> 
>>> Another concept of a type is that it represents a "domain", but that
>>> has a similar lack of acceptance from people that deal with types on a
>>> regular basis.
>>> 
>>> Anyways, if my proposal here has any flaws, I think it's that this
>>> definition of "type" might be too restrictive for some people.  But I
>>> think that that is ok -- you can use another language if you want to
>>> use a different concept of type.
>>> 
>>> Comments?
>>> 
>>> Thanks,
>>> 
>>> --
>>> Raul
>>> ----------------------------------------------------------------------
>>> 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
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to