He has now posted the complete paper on his publications page, including typo fixes and some additional comments at the end.
by Bjarne Stroustrup
From the article:
Here, I will explore, and debunk, five popular myths about C++:
- “To understand C++, you must first learn C”
- “C++ is an Object-Oriented Language”
- “For reliable software, you need Garbage Collection”
- “For efficiency, you must write low-level code”
- “C++ is for large, complicated, programs only”
If you believe in any of these myths, or have colleagues who perpetuate them, this short article is for you...
The 10 sections of this article was posted on isocpp.org in three installments and reposted widely. It attracted a quite varied set of comments. I have now fixed a few typos that were reported. Thanks.
This postscript is my observations on some of the comments posted.
The comments prove – yet again – that the “Myths” paper was needed. People keep repeating the old hairy rationalizations. Unfortunately, many programmers don’t read long papers and dismiss short ones for being incomplete. The unwillingness of many programmers to read a long paper was the reason I released this paper in three separate parts.
This paper is not a research paper carefully outlining every alternative and carefully documenting every detail. I said that right up front:
Each myth requires a long paper or even a book to completely debunk, but my aim here is simply to raise the issues and to briefly state my reasons.
However, many seems to confuse the examples used to illustrate a point with the point itself. Many tried to “debunk the debunking” by changing examples, by changing the constraints on the examples, or by deeming the examples trivial. The examples are small – they have to be to fit into a short paper – but they are not unrepresentative of code found as part of real-world programs.
Many commenters quickly did a shift from the C++11/C++14 that I base my arguments on to some older version. C++14 is not the C++ of the 1980s. It is not what most people were first taught. It is not the C++ that people is presented with in most beginning C++ courses today. It is not what most people see when they look at a large code base. I want to change that. Not being able to do some example that I present in an antique version of C++ or with an outdated compiler is unfortunate, but better versions of the major compiler ship (typically for free) today. I showed no examples of “bleeding edge” code.
The problem of code in older styles is one that every successful programming language must face, so please don’t judge C++ exclusively based on 20-year-old techniques and 10-year old compilers. Look at modern C++ and find ways of getting it into use – many has already. You almost certainly used a program today that was written using C++11. There are C++11 in many steps of the chain between the computer on which I write this and the computer on which you read it.
Quite a few comments were along the lines of “Language X has a feature that does exactly that” and “Library Y in language X does exactly that.” Obviously, if you have a language that provides a simpler solution to the best you can do in C++, with acceptable performance, portability, and tool-chain constraints for what you want to do, use it. However, no language and library is perfect for everything and everybody.
I present examples of general problems and general techniques. Matching a single example in some way is not necessarily significant. My points are general and the examples only simple illustrations. Given a sufficiently good library, just about any programming task can be simple and pleasant. Given a sufficiently constrained task, we can always design a specialized language to be more elegant than a general-purpose one. For example, the asio library I used in §6.1 is a flexible, efficient, general-purpose networking library. For any one task, I could wrap it in a far simpler function (or small set of functions) to make that task significantly more convenient. The code I showed would then be the implementation. My key point in §6.2 is that the C++ library development community could do many programmers a favor by spending a little more tome making simple things simple. For example 99% of the time I prefer
My comments about performance caused quite a stir in places. Many people tried to dismiss them with arguments or mere counter-assertions. I don’t accept performance arguments unsupported by measurements. My comments have been validated by real performance measures in many contexts over years. Many can be found in the literature. My main performance points hold over a wide range of similar examples and scale.
Please note that I assume a modern, standards-conforming C++ implementation. For example, when I talk about the performance of the short-string optimization, I don’t mean a pre-C++11 standard library without that optimization. Also, I take no notice of comments to the effect that C++ facilities such as
std::stringare slow if you don’t use an optimizer – of course they are, but talking about performance of unoptimized code is silly. If you use GCC or Clang use –O2; for Microsoft, use release mode.
I know C and its standard library pretty well. I wrote considerable amounts of C before many of today’s students were even born and contributed significantly to the C language: function prototypes,
inline, declarations in
for-statement, declarations as statements, and more came from my work. I have followed its development and the evolution programming styles in C.
Yes, the C versions of
compose()fails to check
malloc()’s return value. I did ask if you thought I got it right. I did not present production-quality code, and I knew that. Failing to check results is a major source of errors, so my “mistake” failing to check the result of
malloc()was deliberate illustrates a real problem. As in this case, exceptions can often help.
Yes, you could write the C version of
compose()differently using less well known standard-library functions, and yes, you can avoid the use of free store if you let the caller supply a buffer allocated on the stack and let the caller deal with the problem of string arguments that would overflow the buffer. However, such alternatives completely miss the point: it is harder to write such code than the C++ version, and far harder to get it right. Novices get the C++ version right the first time, but not any of the C versions, especially not the C versions that rely on standard-library function not commonly taught to novices.
C++ has been used for demanding embedded systems and critical systems for years, examples are The Mars Rovers (scene analysis and autonomous operations), The F-35s and F-16s (flight controls), and many, many more: http://www.stroustrup.com/applications.html. And, yes, the Orion space capsule is programmed in C++.
Yes, libraries vary in quality and it can be extremely hard to choose from the large selection of libraries beyond the standard. This is a major problem. However, such libraries exist and researching them is often more productive than simply barging ahead and reinventing yet-another wheel.
Unfortunately, C++ Libraries are often not designed for interoperability with other libraries.
Unfortunately, there is not a single place to go to look for C++ Libraries.
I have observed students being taught by the “C first” approach for many years and seen the programs written by such students for decades. I have taught C++ as the first programming language to thousands of students over several years. My claims about the teachability of C++ are based on significant experience, rather than introspection.
C++ is easier to teach to beginners than C because of a better type system and more notational support. There are also fewer tricks and workarounds to learn. Just imagine how you would teach the styles of programming you use in C using C++; C++’s support for those is better.
I would never dream of giving a beginner’s C++ course that
- didn’t include a thorough grounding in memory management, pointers, etc.
- didn’t give the students a look at “plain C”’ and some idea of how to use it
- didn’t present a rationale for the major features
- tried to teach all of C++ and every C++ technique
Similarly, good C teachers do not try to teach all of C and all C techniques to beginners.
http://www.stroustrup.com/programming.html is my answer to the question “How would you teach C++ to beginners?” It works.
For a rather old paper comparing aspects of teachability of C and C++, see B. Stroustrup: Learning Standard C++ as a New Language. C/C++ Users Journal. pp 43-54. May 1999 (www.stroustrup.com/papers.html). Today, I could write the C version a bit better and the C++ version quite a bit better. The examples reflect common styles of the time (and were reviewed by expert C and C++ programmers).
C++ today is ISO Standard C++14, rather than what I described 30 years ago or what your teacher may have taught you 20 years ago. Learn C++11/C++14 as supported by current mainstream compilers and get used to it. It is a far better tool than earlier versions of C++. Similarly, C today is ISO Standard C11, rather than K&R C (though I am not sure if the C compilers today are as close to C11 as the C++ compilers are close to C++14).
I am appalled by much that is taught as “good C++.”
C++ is not (and never were meant to be) an “OOP” language. It is a language that supports OOP, other programming techniques, and combinations of such techniques.
If you are an experienced programmer, I recommend A Tour of C++  as a quick overview of modern C++.