Re: Okay, remember hearing that before, and I understand ...

Post #247,574 by admin 3/11/06 1:19:24 PM Reply	Re: Okay, remember hearing that before, and I understand ... Strict functional programming has no side-effects. What this means is that you can't change anything. You can, however, return something calculated from your inputs. Simplistically, a true functional program has an entry point, and all of the functions within just return a modified version of what's passed in, and the output is the result of calling all of the functions. There are no global variables and no side-effects, and that's why a function will always return the same value given the same inputs. Nothing matters but the parameters that are passed in. Because of this you have to go through some hoops to perform some imperative constructs: looping is done through recursion, for example. // Crappy pseudocode example\nfunction forLoopPositiveIncrementsOnly(int from, int to, int inc, function dowhat)\n{\n if (from > to) return;\n dowhat(from);\n loop(from + inc, to, dowhat);\n} Note that there IS a stack, but you can't modify the values on the stack, just use them. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #247,575 by Another Scott 3/11/06 1:24:33 PM Reply	So you can't have a (pseudo) random number generator?
Post #247,576 by admin 3/11/06 1:34:03 PM Reply	Those don't have side-effects And given the same seed, a pseudo-random number generator will always return the same sequence of "random" numbers. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..."
Post #247,577 by Another Scott 3/11/06 1:52:48 PM Reply	I was too tearse. Some random number generators are more random than others. :-) I'm aware of the limitations of typical pseudo random number generators. I was too tearse in putting that in the Subject. (One of many true random number generators is [link\|http://www.fourmilab.ch/hotbits/\|HotBits].) I wasn't questioning the "side-effects" aspect (I don't think), I was wondering whether it would be possible to hypotheically code a random number generator in a pure functional language given that, apparently, f(x0,t) = y0 for all x0 for a fuctional language, where a random number generator would give f(x0,t) = y(x0,t,gamma) (where gamma is some truly radom parameter). (If I didn't mess the notation up.) Presumably the code would talk to a radioactivity sensor or something similar that was truly random. I would think that it must be possible, but wonder how it would be simply explained and whether it violates the premise of functional programming. Thanks. Cheers, Scott.
Post #247,578 by admin 3/11/06 2:00:46 PM 3/11/06 2:02:24 PM Reply	For pure functional programming, no Because the program won't be provable for a given set of inputs. Unless you count the external random number as one of the inputs, I suppose. Regards, -scott anderson "Welcome to Rivendell, Mr. Anderson..." Edited by admin March 11, 2006, 02:02:24 PM EST Expand All History
Post #247,587 by ChrisR 3/11/06 4:01:46 PM Reply	You can if you thread the state through the functions... ...and hence the introduction of monads.

Welcome to IWETHEY!