"Christof Ressi" christof.ressi@gmx.at writes:
I have - and traversing large lists of scalars by pointer can be veeeery slow.
yeahhhh, i figured that could be the case =P
you could hide the fact that it's a linked list from the user but I don't know if that's a good thing to do. It doesn't seem transparent to me ("why is accessing the 1000th elements slower than the 1st element?").
I suppose you could tell the user via documentation, but then you'd have go into what a linked list is and why linked list indexing is O(n) versus array O(1) to explain the speed difference and yeah, I see your point haha. And it's probably not good either to have a layer of abstraction that isn't totally necessary either.
if you don't use the graphical representation, you can just as well work with data structure arrays and get fast random access. there is one little but unfortunate drawback compared to using lists of scalars: drawn instances of array elements don't respond to mouse events, making them more or less useless as UI elements. if you click/select an element, you only get a pointer to the array, not to the element(s). if you click on the canvas, you get pointers to the array + all array elements (for whatever reason). The right behaviour IMHO would be that clicking/selecting array elements gives you pointers to the array *and* the clicked/selected element(s). clicking on the empty canvas shouldn't trigger any mouse events!
Apart from that, data structure arrays are quite easy to handle and much more efficient than linked lists of scalars
ah, that's interesting. admittedly, i haven't used pd structs that much in my pd work as of yet but the stuff i want to do seems to be increasingly lending itself to that (or a similar) sort of paradigm so i'll keep that in mind!
Derek