3 ms·
Interesting, but one could go further with this: there still seems to be the assumption that learning OOP concepts is indispensable and necessary in the long ru
by onan_barbarian 16y ago
Interesting, but one could go further with this: there still seems to be the assumption that learning OOP concepts is indispensable and necessary in the long run.
This may or may not be true depending on the application area, language and intent of the programmer. Some languages and frameworks make it impossible to get anything done unless you know how to subclass things; at the other extreme you've got Stephanov-inflected C++.
Arguably you would have people learn at least a little more about algorithms and data structures of greater complexity before dropping the OO-hammer. I still have moments where mid-way into astronauting up a class hierarchy I realize that all of this could be done more clearly in 10 lines of STL/Boost (and yes, I'm aware of the problems of that style of coding, too).
- nickik 16y agoI agree but STL/Boost hard tiny bit to hard. Teach that stuff with Scheme, python or something that isn't <insert what you hate about C++>
- onan_barbarian 16y agoYou misunderstand. I'm merely using STL/Boost as one example of advanced non-OO programming. Many other such paths to sophisticated programming exist that don't go through OO, including the ones you name. I use STL/Boost in this case not because I think they're wonderful but because they're realistically what I'd have to use, as our codebase is mostly C/C++.