C and C++ Background

The creation of C is closely associated with the early days of Unix. Bell Labs developed Unix out of an earlier project called Multics. The first version of Unix ran on PDP-7 microcomputer and funding was given to move it to PDP-11. Dennis Ritchie was a key member on this project and set about creating a language that could help him develop Unix while minimizing the amount of assembly language he had to write. Most of the code up to that point was expressed in assembly language which was error prone and obviously non portable.

Ritchie developed C so that he could write code in terms of variables, expressions, loops, functions etc. and use a compiler to translate C code into machine code. The generated code ran almost as fast as hand written assembly and was more portable since only the compiler had to be changed in order to support a new architecture. C itself was influenced by B (hence why it was called C), which itself was influenced by BCPL.

In 1978 C was formalised into a defacto standard called K&R C, named after Brian Kernighan & Dennis Ritche who published the standard as a book.

Over time the use of C became more widespread and compilers such as Turbo C, Lattice C, Microsoft C popularized C on other operating systems including personal computers.

C later became an ANSI standard, C89. A further standard followed with C99 and C is still under review and development.

Some functionality that was introduced in C++ has also found its way back into C standards. For example, the // style single-line comment and variable declaration rules in blocks.

C++ first appeared in 1983 as C with classes. It was invented by Bjarne Stroustrop as a way to imbue C with Simula-like features. Simula is a language that allowed concepts such as objects, classes and inheritance to be expressed in code and as its name suggests was created for running simulations. However it was considered too slow for systems programming and so something that combined speed of C with object oriented concepts was highly desirable.

C++ became popular in its own right and outgrew the limitations of cfront preprocessor to become supported by compilers in its own right. Thus toolchains such as Microsoft Visual C++, GCC, Clang etc. support both languages. Some toolchains have also been given to favouring C++ over C, for example Microsoft’s compiler has been very slow to implement C99.

Object oriented programming has mostly been used in higher level software - applications, games, simulations and mathematical work.

C++ has also become formalised standards with C++98, C++03, C++11 and so on.

Modern C++

C++11 onwards is a distinctly different beast from earlier iterations and strives to add functionality that if used correctly can eliminate a lot of issues that will be discussed later on:

• auto keyword
• move semantics (i.e. moving data ownership of data from one variable to another)
• rvalue references
• perfect forwarding

However it is worth noting that since many of these things are late additions to C++. Things like move semantics must be explicitly used and have implications that are not an issue for Rust where they have been part of the language since early on.

While C++ grew out of C and has developed alongside it, it is not true to say C++ is a superset of C. Rather it is mostly a superset. There are differences such as keywords and headers that C recognizes that C++ does not.

C++ has function overloading and classes and uses name mangling to disambiguate overloaded functions. But in practice it is possible to write C as a subset of C++ and compile the two into the same executable. Most real-world C code could be called C++ without classes.

C++ tends to be more popular with applications level programming. Part of the reason C++ hasn’t found itself in the lower layers is the perception that exception handling, name mangling, linking and issues of that nature add unwanted complexity or that somehow the generated code is less efficient. Arguments have been made that this is not the case, but the perception still remains.

C still tends to be more popular in low level systems programming. Components such as the Linux kernel are pure C with some assembly. Many popular open source libraries such as sqlite3 are also written in C.

Objective-C is another C derived language that added objects and classes. Unlike C++, Objective-C behaves as a strict superset of C.

The language was developed in the 1980s and was popularized in the NeXTSTEP operating system and later in Apple’s OS X and iOS. It hasn’t gained much popularity outside of those platforms but the success of the iPhone has ensured it has a sizeable developer base of its own. It is also well supported by the GCC and Clang toolchains. Apple has begun to deprecate Objective-C in favour of Swift which is a modern high level language similar in some respects to Rust but more application focussed.

Objective-C is strongly influenced by Smalltalk (as opposed to Simula in C++) and so code works somewhat differently than C++.

Notionally code calls objects by sending them a message. An object defines an interface specifying what messages it accepts and an implementation that binds those messages to code. The caller code sends a message to call a method. Objects can also receive dynamic messages, i.e. ones not defined by their interfaces, so they can do certain tasks such as intercepting and forwarding messages. In addition an object can ignore a message or not implement it without it being considered an error. In a broad sense, an ObjC message and a C++ method are or more or less analogous in functionality.

These are the major revisions of C and C++