In article <a995ot$9ub$[EMAIL PROTECTED]>, [EMAIL PROTECTED] (Ton Hospel) writes: > Ok, this one is now finished too. > There is no post-mortem link yet, but you can find it at > http://la.pm.org/cgi-bin/PGAS/post_mortem.cgi?id=1 > > It's won by Spifff with an entry that the judges of tpr3 > should take notice off. It probably means the rule > "must run in finite time" should be amended to something > like "expectation value of the runtime should be less than > a week" >
Ps, notice I put the "expected" there to allow cases like this shuffle: @_=<>;print+delete$_[rand@_]while@_ which is not guaranteed to finish, but the expectation value (well-defined math term) is finite and not even big. I'd prefer it if the rules allow solutions like this. Also, to not depend on the speed of particular computers, and allow the judges some leeway: "The program is expected to finish in a reasonable time" (which is nicely vague, has the "expected" term to allow monto-carlo style progams, but should allow to drop real abusers) Then there is the $^T that's sometimes used and makes solutions depend on when they run. I propose to formulate the rules so that a program only has to be valid during the period that the contest runs. 1970 or whatever your computers epoch is, is long gone, so $^T tends to be "big"
