[Date Index] [Thread Index] [Author Index]
RE: Problem with Times. Bug ????
Dave Withoff of WRI writes: >The Orderless attribute only affects the treatment of commutativity in >the main evaluator and the pattern matcher. Individual functions are >still free to implement whatever rules they want. There are thousands >of places in the kernel where multiplication is assumed to be commutative, >regardless of whether or not the Orderless attribute is set. It would >be pretty hard to do otherwise -- many basic algorithms make assumptions >such as these about elementary arithmetic operations. This deserves a translation: "Don't fool with the Times function in this way. You can't make it work unless you know exactly what your're doing. Even if you do, it will probably still not work. There will be strange errors that seem to have nothing to do with your code. It is better to write your own operation from scratch." Although given the number of people who get caught up in this, WRI really ought to provide some sort of facility for those wishing to define new operations. Richard Mercer Wright State University rmercer at desire.wright.edu