Re: APL & Programming Concepts

Randy MacDonald wrote:
> phil chastney <phil.hates.spam@[EMAIL PROTECTED]
> wrote in
> news:HaoWj.226457$XH2.46775@[EMAIL PROTECTED]
 
> 
>> Randy MacDonald wrote:
>>> <snip>
>>>
>>> No APL vendor is going to know all the needed primitives in advance, 
>>> so that expectation is kind of excessive.
>> nobody asked the APL vendors to know all the needed primitives in
>> advance 
> 
> Perhaps I'm misinterpreting your earlier statement:
>> I don't want to use named cover functions for rational multiplication,
>> any more than I want to use them for complex multiplication
> 
> I see it as saying that one wants APL primitives to handle one's pet
> area of functionality.  Once I developed 'new' and 'to' functions,  I no
> longer needed to wait for the whims of the vendors: 
> 
> Perhaps I should have said:  no APL vendor is going to have primitives
> for everyones pet functionality, because once one person asks for it,
> everyone will feel they have the right.  That doesn't scale. 

but again, nobody asked the APL vendors to provide primitives for 
everyone's pet functionality

all the vendors need to do is provide the extendability

some people might then like to define their own types -- i.e, 
user-defined types, as opposed to primitive types -- without waiting on 
the whims of the vendors

of course, once somebody has a set of routines which define, say, 
rationals or complex numbers, and routines which cover a useful subset 
of the primitives, it would be nice if that could be made publicly 
available as a library

then the rest of us (if we wanted to, of course) could plug it into our 
APL, and bingo! we could all have APL with rationals, say, or complex 
numbers, using special symbols

and the half dozen people who want quaternions could move forward from 
there (but the rest of us wouldn't bother plugging that in)

as an aside: previous debates about APL's past and future have shown a 
distinct split in the APL community between those who would be prepared 
to give up special symbols, in the interests of wider acceptance, and 
those who would definitely prefer to keep them

I definitely prefer special symbols  --  if I wanted to write my 
arithmetic out in words I'd use Cobol

regards   . . .   /phil