Yes, there is a duality but I can't see any purpose in mapping that into the same precedence rule.
No, my notions of "main" and "change" aren't arbitrary; among other things they have ground in the behavior of realloc(3), which makes a good efficiency reason for the distinction I make. It's true that it isn't connected to the logical operations per se, but the operators that implements the operations should nevertheless take it into account, since the op= forms in fact exist and makes it possible to optimize many instances to destructive operations.
/ Martin Stjernholm, Roxen IS
Previous text:
2002-12-03 23:37: Subject: `| and order on arrays
It's not the same thing, but it's quite dual. In one case, it's a question of precedence of values for the same index, and in the other it's a question of precedence of index for the same value.
Since the operations are not destructive, your notions of "change" and "main" are quite arbitrary and not inherent in the operations themselves.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)