User talk:Comps

Rule 90

Welcome to RC! I noticed you requested the Rule 90 game earlier. I suggest modifying the rules in One-dimensional cellular automata to do what you need. It seems that Rule 90 is just a variant of the "game" described there. --Mwn3d 20:00, 18 April 2011 (UTC)

I think that One-dimensional cellular automata corresponds to Rule 104. Rule 90 would be a different task, I think? Or, an alternative would be to suggest that One-dimensional cellular automata be co-opted to become a general implementation, covering all rules (but this would mean that all current implementations would need to be updated). --Rdm 15:49, 19 April 2011 (UTC)

I think Rule 104 and Rule 90 are too similar to warrant separate tasks. They would be exactly the same except for the table of rules at the end of the description. The only code changes would need to be with the code that defines how to react to different numbers of neighbors. --Mwn3d 16:01, 19 April 2011 (UTC)

They are structured similarly. However, in the general case, Rule 90 needs more robust "end state testing" than Rule 104 needs. (You can tell when you have reached an end state in rule 104 when two adjacent generations are identical -- when nothing changes. But this may never happen with Rule 90.) In addition, Rule 104 can be defined in terms of the sum of the three cells, which is not the case for Rule 90. So I think Rule 90 is a more general task than Rule 104. And we certainly have other cases where extremely similar processes have multiple tasks (sorting, for example, or Counting in Factors vs. Prime decomposition. Anyways, yes, we could take over One-dimensional cellular automata and make it become a generic "Rule n" task -- but I am not sure I like invalidating all the existing implementations of that task. --Rdm 16:36, 19 April 2011 (UTC)