I don't think I agree with this. As nearly everyone has pointed out, 99.9% of the cases are not order-dependent, so that all orders give the same results. The non-onehot uses of unique, using your terminology, are "illegal" (a poor choice of terms) not because they require non-deterministic behavior, which they don't, but rather simply because they are not one-hot. Period. That is, as Cliff as pointed out, unique is an assertion of one-hot. A non-onehot use is simply a violation of that assertion. > We say that non-onehot uses of unique are "illegal" because, > formally > speaking, they require non-deterministic simulator behavior, > which > cannot be delivered. The simulator "warns" when it must > discard an > alternative simulation control flow. ShalomReceived on Mon Jan 23 07:05:20 2006
This archive was generated by hypermail 2.1.8 : Mon Jan 23 2006 - 07:06:17 PST