al_batuul
11/20/2007 2:10:00 AM
Note: parts of this message were removed by the gateway to make it a legal Usenet post.
One step is needed before the data can define itself is to design a class for training data, to learn about all different data types. Then it is possible for the class to discover and classify its own type by itself.
I think the number of iterations should be defined within the type algorith and associated to the purpose of the open-ended class,
I would like to see some serious progress in developing this class;
al_batuul
Yossef Mendelssohn <ymendel@pobox.com> wrote: On Nov 19, 6:54 am, "Todd Benson" wrote:
> On Nov 19, 2007 6:44 AM, al_batuul wrote:
>
> > Dear Clifford,
>
> > Defining an open-end class is a great idea; the rang from 1..nil or 0 to nil could help now, I suggest you add different types of parameters to the class such as time.
>
> > all the best,
>
> Yes, would it help you really?
>
> Let's make all data define itself. Let's make stuff work for now.
> I'm not convinced that 1 to (what's supposed to be) some object of
> NilClass will help anybody anywhere.
>
This subject has come up more than once. I've worked on it myself, and
found a passable (if incomplete) solution.
What I don't understand is why so many people find an open-ended range
to be pointless. That is to say I do understand why --- because
they're set on iterating over the range --- but I don't understand why
they consider that to be the only use of a range.
Everyone I've seen who's interested in this (myself included) is using
these ranges for another reason, to indicate a range of acceptable (or
maybe even unacceptable values). A range like 5..Infinity is not
intended for iteration, but for comparison.
--
-yossef
---------------------------------
Be a better pen pal. Text or chat with friends inside Yahoo! Mail. See how.