Although I have no doubt that Roger is fully aware of the Master’s advice "It 
is very important, very important to listen to your customers.  It is even more 
important to disregard what they say and do the right thing" he appreciates 
some comments and suggestions from users, yours included, as threads of this 
forum testify.
 
I would certainly prefer, for instance, "using GPUs as numeric coprocessors" 
but I do not think the efforts involved are comparable and one should take into 
account the bang for the buck.  Yet, for me the rank definition issue is 
admittedly a matter of aesthetics akin to removing a small eyesore from a 
beautiful landscape.  After all, from a practical perspective, one can easily 
code a conjunction having the proposed behavior without the need to change 
anything:

rank=. 2 : 0
  u " (v y) y
:
  x u " (x v y) y
)

Incidentally, how could you write (rank) above tacitly using the long gone 
tacit facilities (I was never close to be fluent because there were too many 
combinations for me to remember)?


________________________________________
From: [email protected] [[email protected]] On 
Behalf Of Henry Rich [[email protected]]
Sent: Wednesday, November 17, 2010 8:30 PM
To: Programming forum
Subject: Re: [Jprogramming] Atop u...@v with v of negative monadic rank

I think Roger has the best view & so I would like him to do what he
thinks is right.  Fortunately, that's what's gonna happen no matter what
we say here.



But I am interested in things like:

Support for semiduals (like &.: but applying v to only one operand and
the result)

Using multiple cores effectively in a single J program

Using GPUs as numeric coprocessors

Taking advantage of SSE3 instruction set

Hand-coding +/ . * using SSE3 to get the ultimate perf in matrix
multiply (I have offered to code the inner loops for this)

perf improvements in ;@:(f&.>)

A primitive like -: that checks the type of empties


...and, I'd like to get the tacit language back, just as a monument to
Ken's brilliance.

Henry Rich


On 11/17/2010 7:58 PM, Dan Bron wrote:
> Obvious follow-up question:  what would you like to spend it on?
>
> -Dan
>
>
> Please excuse typos; composed on a handheld device.
>
> -----Original Message-----
> From: Henry Rich<[email protected]>
> Sender: [email protected]
> Date: Wed, 17 Nov 2010 18:46:49
> To: Programming forum<[email protected]>
> Reply-To: Programming forum<[email protected]>
> Subject: Re: [Jprogramming] Atop u...@v with v of negative monadic rank
>
> Reality check: the most precious commodity in the community of J users
> is Roger's time.  Is this what you want it spent on?
>
> Not me.
>
> Henry Rich
>
> On 11/17/2010 5:25 PM, Jose Mario Quintana wrote:
>> I just want to add my name to the list of those who would support the 
>> interpretation change of (f " g) to
>>
>> f " g y<->    f "( g y) y    ( and x f " g y<->   x f " (x g y) y )
>>
>> from the rarely used (or useless, depending on your point of view) current 
>> definition.
>>
>> ----------------------------------------------------------------------
>> 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


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

Reply via email to