On Tue, Aug 4, 2009 at 6:48 PM, Kip Murray<[email protected]> wrote:
> As a start on Raul's problem below, would you be willing to use the following
> for-this-problem definitions?
>
> A "sequence" is a list of boxes.  A "term" of a sequence is the open of one of
> its boxes.
>
> ...

I think it would be better to not focus on boxing in this manner.
Instead, I propose that you define things in terms of lists of items.
That way many example sets could be handled as literal or numeric
arrays. So defined you can have sets where the members are
multidimensional arrays (if all those members have the same shape and
datatype), without boxing. That flexibility might not be valuable, but
it may emphasize how much can be done without managing boxing.

There are times where boxing is a natural part of the solution, such
as producing the power set. I'm only urging that you not limit the
representation of all sets to lists of boxes, as the J concept of
"item" is so handy a fit to your purpose.

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

Reply via email to