Agreed in toto. As usual; almost all the time (some few exception (grin)).
What is needed is a systematic approach to studying such things with
some form of metrification.
Lacking that it becomes a matter of taste and test of time.
I _assert_ there are (probably) better language packings, and I also
_intuit_ there is room for moreso in J.
BUT: I have not provided them and as yet I do not intuit enough to
approach this constructively.
HOWEVER: the J community as a whole _does_ stand a (far better) chance
(than one person) of getting this
_correct_
when suggesting such (against the level of experience and correctness
already evident from RH);
it's just prkaktka d--n difficult to do so.
The flow of the levle of experience and inner "athema" is very, very high.
The good or perhaps _best_ thing about J is that is is easy to make your
own.
With some few exceptions. My most basic hmph is that all syntactic
classes should be modifiable in full;
but this in turn has it's problems.
Only a systematic look will do: and that is nto going to be trivial.
Roger Hui wrote:
But is there an organized collection of these nits somewhere?
You can write an essay about it in the J wiki.
Another one:
The foreign m!:n would be slightly more concise and easier
to use if !: were an adverb with the encoding being (n+100*m)
or perhaps (n+10*m). Or perhaps use : and word formation
instead of !: and adv/conj:
old new
3!:0 30!: or 30:
3!:1 31!: or 31:
3!:2 32!: or 32:
Another one:
Since the dyad u/ is just u"(lu,_) , perhaps it should be
assigned a different, more useful meaning. e.g.
x u/ y
x if 0=#y
(0{y) u (1{y) u ... u (_2{y) u (_1{y) u x if 0<#y
That is, > u&.>/ (<"_1 y),<x if 0<#y .
Another one:
Array semantics but "LISP syntax". e.g.
((" * 1) (i.2 3) 3 4 5) instead of (i.2 3)*"1 (3 4 5)
etc. Once you start this game the possibilities
are endless.
----- Original Message -----
From: Raul Miller <[EMAIL PROTECTED]>
Date: Wednesday, September 5, 2007 9:52
Subject: [Jgeneral] subjunctive J -- @ v. @: and & v. &:
To: General forum <[email protected]>
It seems to me that a philosophy of J might be better served if
@ and @: had their meanings swapped. Likewise for & and &:
This is because (except in cases altered by special code), @:
tends to be faster than @ (and I believe the same holds for
& and &:). The main corresponding philosophy is that shorter
code should tend to be faster than longer code. A secondary
philosophy is that the trailing : would a visual pun suggesting
treatment of small pieces.
On the flip side, of course, is the philosophy of supporting
existing users -- if this were changed, all sorts of code would
subtly break. A related issue is that there may be a
variety of
other "would be nice, or at least interesting" changes of a
similar character which are currently unidentified.
I'm not necessarily going anywhere with this -- I classify
this issue in the "worth noting, but not an action item" bin.
It's been on my mind, so I figured I'd write this down.
But is there an organized collection of these nits somewhere?
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
