You still need to know what your library's performance is like under various conditions - large or small numbers of items, whether comparisons are cheap or expensive, how much extra memory it takes up, etc. If you don't understand the library, how are you going to be able to use it, except in relatively trivial cases?
I don't debate library usage (reuse!), but I teach mostly theory, with code snippets to present algorithms concisely.
There are many tools to solve a problem, and I want to educate on a different layer. For instance, in sorting alone there are many sorting algorithms, I don't expect someone to interrupt me with "just use quicksort!" when I'm teaching bubblesort. There are lessons to learn in both, but it's a bit presumptuous (or at least, there are less smug ways) to assume ignorance, and not intention.
Sorry - that was sort of my point, but I didn't put it very well. Your student likely isn't taking any of that into account. One way to get your point across is to ask them a few pointed questions about (eg.) the data set that they're sorting. If it's relatively small, then the overhead of quicksort might not be worth it. If it's already mostly sorted, then bubblesort is potentially better, and so on.
