[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

:zz: Cell-identity recognition

Well, actually, considering what display boards do already,
 the problem of having cell-identity recognition done
 down the chain from the other functions might be
 a very good one.  I'm sure it *could* be done deep in the
 display routine; it's just not feasible at this moment.


At 11:30 AM 7/27/98 -0400, you wrote:
>At 10:28 PM 7/25/98 +0900, you wrote:
>>>   However, since there can be
>>>   an infinite number of such equivalences, and connecting some visually
>>>   would conflict with connecting others, you'd need a rule to arbitrate
>>>   (probably part of the same algorithm that decides which parts
>>>   of the infinite display to update first).  
>>Right!  As long as we're omniscient about the equivs,
>> anyway, which is a big sticking point.
>I guess what I was saying was, since you obviously can't be omniscient,
>you need a "heuristic" that finds and shows some equivalences.  That 
>means that it doesn't find others, and also that if it chooses to show one
>equivalence by connecting some cells in a loop, then there may be other
>equivalences that are made impossible to show (in this unmodified 2D
>grid visualization) because of that choice.  Knots.  But you just 
>invent & try simple heuristics until you've got one that you kinda like, 
>and live with it.  (Just thought of the possibility of letting the user
> --Steve
Theodor Holm Nelson, Visiting Professor of Environmental Information,
 Keio University, Shonan Fujisawa Campus, Fujisawa, Japan _____
http://www.sfc.keio.ac.jp/~ted/   ______
Permanent Address: Project Xanadu, 3020 Bridgeway #295, Sausalito CA 94965
_____  World Wide Fax +1-415-332-0136   World Wide Voicemail:
+1-415-331-4422 _____ PERMANENT E-MAIL: ted@xxxxxxxxxx