A clarification on my last post on the indexing for static code: The indexing is done for one argument, but indexing code for all arguments where indexing is useful will be generated, so if you have something like f(1, a). f(2, a). f(1, c). and you call the goal f(1, X), indexing will be done on the first argument, so f(2,a) would be filtered out and not tried. If you call the goal f(X, c), indexing will be done on the second argument, so f(1, c) will be the only clause that will be tried. In outline, the indexing code looks like if argument1 is an integer, index on argument1 if argument2 is an integer, index on argument2 If you call f/2 with both arguments instantiated, e.g. f(2,a), the indexing is still done with one argument. If the indexing is on the first argument, f(2,a) will be the only clause tried, but if the indexing is on the second argument, then f(1,a) and f(2,a) will be tried. What the compiler decides by heuristic is if the indexing is done on the first or second argument when both arguments in the goal is instantiated, i.e. to decide between if argument1 is an integer, index on argument1 else if argument2 is an integer, index on argument2 if argument2 is an integer, index on argument2 else if argument1 is an integer, index on argument1 Cheers, KishReceived on Wed Sep 12 2012 - 23:37:15 CEST
This archive was generated by hypermail 2.2.0 : Sun Sep 16 2012 - 06:14:57 CEST