So you say “> is as > does” This is reminiscent of duck typing. I agree one may view things that way.
But what does your correct observation “Non-numeric domains are irrelevant in this context.” mean? Does it mean we have to view any function as its restriction to the numeric domain? That, combined with the duck-view is consistent with the way OP wants derivation to work. And again, I agree this can be done. Personally, I’d be more restrictive and say: IMO, there are functions meant to work on numric arguments and there are functions that aren’t but still support them. And I’d consider the latter ones unfit for derivation. In the end, people want to get things done and using a documented hack is not in itself a bad thing. It’s a matter of opinions – and since my view is the more restrictive one, it might actually be a bad choice from a pragmatic point of view which I agree might be favoured over a “puristic” or whatever you want to call it point of view like mine. I keep my opinion but maybe it’s better for the language and its user base to consider any function to mean its de facto restriction to the numeric domain when it comes to derivation. (de facto as opposed to my prejudice about its intended semantics) Thanks for making me rethink Am 17.01.21 um 06:36 schrieb Raul Miller: > I disagree. > > The derivative of > is exactly the same as the derivative of ] > > Both work on boxed arguments. > > Both produce the same result on numeric arguments. > > The derivative of ] is the same as the derivative of -@- > > -@- does not work on boxed arguments, but has the same result on > numeric arguments as ] does. > > This is simple, straightforward mathematics. > > Non-numeric domains are irrelevant in this context. > > Note also that 4: accepts boxed arguments, and has a derivative. We > ignore that we cannot take the difference between two boxes in that > example. > > Thanks, > -- ---------------------- mail written using NEO neo-layout.org ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
