|
Bob Armstrong Feb 25, 6:43 pm
Newsgroups: comp.lang.apl
Date: 25 Feb 2005 18:43:49 -0800
Local: Fri, Feb 25 2005 6:43 pm
Subject: Re: Function Valences
Hi Jim ,
What are you up to these days ?
TaliesinSoft wrote:
> As I see it there are eight potential valences for an APL function,
> either primitive or defined.
I wrote a rejected paper back in the `80s which laid out the same
table . I , also , have always rued the lack of omni-adic words .
I should be able to make a one word sentence , a 2 word sentence
or a 3 or more word sentence with a word knowing its meaning in
context . My example was always
LIBNAMES | current library
LIBNAMES LIB | some other lib .
I'm finally byting the bull and creating my own , simplest ever APL
from the FORTH up .
It just has lists of lists and dictionaries , ala K .
Assignment is just the storing of a symbol - value pair in
the dictionary .
I am becoming inclined to use FORTH's ` [ tick ] function to
mark the following word as a symbol rather than K's `symbol
but in any case assignment will be a simple function taking
a symbol and a value . For example :
` minus <- { la - ra } | or | `minus <- { la - ra }
depending on what I decide . White space IS the prime
delimiter so making words with names like " <- " is no
problem .
I am now immersed in adapting the windows examples in the
FORTH so I can get myself into a CoSy editing/computing
environment . I then will begin seriously creating vocabulary .
The syntax will be very much determined by the words themselves
built upon the basic "read it" "do it" FORTH loop but most
words stacking themselves til the end of the expression is
encountered .
I'll be looking for people interested in being
alpha-user/LanguageSmiths , investors , end-users , etc within
the next couple of months . I definitely intend for this to be
a commercial product - an ultimate intelligent user's computing
environment .
I give only the tiniest damn about speed ; I just want the
ultimate mental and everyday business crutch . For instance ,
I've implemented modulo indexing . If anybody needs to make
something faster , they can just go down to the FORTH level
and make some words that more directly implement what they
need to do .
I have to thank Ray Cannon for inviting me down to Naples
where I seriously started to work on this ultimate , in my
lifetime , CoSy .
I invite anybody interested in being kept up to date on the progress of
this project to go to my site and drop me an email .
--
Bob Armstrong -- http://CoSy.com -- 212-285-1864
NoteComputing Environment : http://CoSy.com/CoSy/
WTC vision : http://CoSy.com/CoSy/ConicAllConnect/
Liberty : http://cosy.com/Liberty/
--
> The same can be said of operators. If there was a
> system function query as to the valence of a function one might have
the
> following values returned:
> 0 0 0 Non value returning niladic
> 0 0 1 Non value returning right monadic (dextral monadic)
> 0 1 0 Non value returning left monadic (sinistral monadic)
> 0 1 1 Non value returning dyadic
> 1 0 0 Value returning niladic
> 1 0 1 Value returning right monadic
> 1 1 0 Value returning left monadic
> 1 1 1 Value returning dyadic
> All of these valences can concurrently exist within the following
rules:
> a) Within an expression determination of valence is made from right
to left.
> b) The valence with the highest value takes precedence.
> c) The entirety of the expression is examined under rules a and b to
> determine the active valences of each function or operator.
> Furthermore, "full valence potential" can be implemented in a manner that
> preserves the current conventions of argument association.
> Unfortunately function assignment as currently defined (an major mistake in
> my opinion) prevents the opening up of APL to what I'll dub "full valence
> potential."
> And..... my motivation for commenting and, yes, grumping comes from my yet
> again wishing I could have such as a function that can be called with
> or without arguments.
> -- James L. Ryan -- TaliesinSoft
|
cursor hover comments