On Fri, Aug 14, 2009 at 11:51 PM, Kip Murray<[email protected]> wrote:
> I AGREE as you will see. I prefer a set to be either 0$<1 or a list of boxes
> without duplicates.
Let us imagine that you have unsorted lists representing sets.
Let us also imagine that you have a set of sequences of sets of sequences
of sets.
How can you determine if one such set is a subset of another such set?
If your representation of sets had used sorted lists, the answer would
be the same as for any other question involving subsets. But if you
use unsorted lists, you must introduce special code which knows
about this specific application domain, to sort your sets (level 4, 2 and 0)
just before you make your comparison.
Note also that if you use realset on the sequences (level 3 and 1)
that you will sometimes be treating sequences as identical which
were different.
The sequence 1 0 1 0 1 is different from the sequence 0 1 0 1 0.
Likewise, the sequence {{}}, {}, {{}}, {}, {{}} is different from
the sequence {}, {{}}, {}, {{}}, {}. And, so on...
>> Nor does it define an API.
> >
>> It's sort of half-way inbetween.
>
> WHAT'S AN API? (I'm serious, I do not know.)
http://en.wikipedia.org/wiki/Api
> Following a plan of Fraser's for a much broader context (set = any array),
I do not see the benefit in this form of generality.
I mean, yes, I can see that the array ?100#2 can represent a set, and
so can i.2 3 4 5 6, but I do not see any value in introducing complexities
so that I can manipulate them with the same primitives.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
