C may give you a reason to do it - because it doesn't include an implementation by default. But that doesn't mean it's the ideal language to use when learning these things. The same algorithms and data structures can be implemented using modern C++, which can be far more intuitive for new programmers.Originally Posted by whiteflags
What can this strange device be?
When I touch it, it gives forth a sound
It's got wires that vibrate and give music
What can this thing be that I found?
Experience (from what I've gathered from people who teach C++; source: Cppcon) has shown that people tend to struggle less if you let them believe data structures are magical when you do intro into programming. When you've taught students the basics, THEN and only then should you venture into details. It means less scratching heads, better learning experience, students generally more happy and less giving up because it's too hard.
The idea is teach 'em the basics first, then teach 'em the details in the more advanced follow-up courses. C can't do that because it forces you to focus on the details first because you have to explain pointers, how to implement and use a linked list, why we null-terminate strings and how we need to allocate space for strings, etc, etc. In modern languages such as Java or C# or C++, this is all done for you. It's magical. You don't need to learn the details until later when you're more ready for them. And that's why I disagree C is a good introductory language.
If you want to learn C, do it as an advanced course, not an introductory one.
Adding a little to the discussion -- I agree with using higher-level languages first, but only if the implications of using the cookie-cutter data structures are explained. Example: you can be super inefficient if you use lists and don't know the complexity/big O time for the different operations. You don't want to learn the wrong way and then learn later when you learn C that the way you've been doing it all along is super inefficient. That's what I did by learning Lisp first....
You guys make good points
I don't consider C++ and C much different in their ability to expose the type of fundamental knowledge required to make a good degree in CS. In a tutoring environment, the STL can be completely ignored and the only thing distinguishing both languages will be their semantics. I would place both C++ and C and equally valuable tools for learning programming. Except...
C may have an edge over C++ because it more quickly and easily exposes the architectural weakness of our computers. Weaknesses that we must learn to live with. Dangerous stuff like the fragment above, open up many opportunities for a deeper understanding of how our computer languages work (*). But C++ isn't far behind.Code:char *str = "A string.";
In addition, what makes me think that I would prefer a CS course based on C is the simplicity of the language, that allows for a larger part of the learning to be focused on the scientific aspect of programming which is the core of a CS course, rather than the engineering aspect which is represented by programming languages and their specifications. C is just easier to learn and understand. The language quickly gets off the way. It is however, like C++, an hard language to master. But that concerns only the task of coding functional and well-formed programs. Which is, generally speaking, not of interest to a CS course.
---
(*) to clarify, code like the above better represent the price we pay for abstraction. It's not really teaching us about how our computers work, but why higher level computer languages come with a cost. That cost is however not a failure of the computer language, but a result of weaknesses in computer architecture that force the hand of the creators of computer languages.
Last edited by Mario F.; 10-07-2016 at 05:29 AM.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
