C++ Crash Course: a fast-Paced Introduction

Chapter 21: Writing Applications
This final chapter rounds out the 
book with a discussion of several important topics. You’ll learn about 
program support facilities that allow you to hook into the application life 
cycle. You’ll also learn about Boost ProgramOptions, a library that makes 
writing console applications that accept user input straightforward.
N O T E
 
Visit the companion site https://ccc.codes/ to access the code listings contained in 
this book.

---

---

A N O V E R T U R E T O 
C P R O G R A M M E R S
This preface is meant for experienced C 
programmers who are considering whether 
or not to read this book. Non–C program-
mers are welcome to skip this prelude. 
Bjarne Stroustrup developed C++ from the C programming language. 
Although C++ isn’t completely compatible with C, well-written C programs 
are often also valid C++ programs. Case in point, every example in The C 
Programming Language by Brian Kernighan and Dennis Ritchie is a legal 
C++ program. 
One primary reason for C’s ubiquity in the system-programming com-
munity is that C allows programmers to write at a higher level of abstraction 
than assembly programming does. This tends to produce clearer, less error-
prone, and more maintainable code.
Generally, system programmers aren’t willing to pay overhead for pro-
gramming convenience, so C adheres to the zero-overhead principle: what 
you don’t use, you don’t pay for. The strong type system is a prime example of 
a zero-overhead abstraction. It’s used only at compile time to check for pro-
gram correctness. After compile time, the types will have disappeared, and 
the emitted assembly code will show no trace of the type system.
A
rthur
 D
ent
: What’s the matter with him?
h
ig
 h
urtenflurst
: His feet are the wrong size for his shoes.
—Douglas Adams, The Hitchhiker’s Guide
to the Galaxy, “Fit the Eleventh
”

---

xxxviii
An Overture to C Programmers
As a descendant of C, C++ also takes zero-overhead abstraction and 
direct mapping to hardware very seriously. This commitment goes beyond 
just the C language features that C++ supports. Everything that C++ builds 
on top of C, including new language features, upholds these principles, and 
departures from either are made very deliberately. In fact, some C++ features 
incur even less overhead than corresponding C code. The 
constexpr
keyword 
is one such example. It instructs the compiler to evaluate the expression at 
compile time (if possible), as shown in the program in Listing 1.
#include
constexpr int isqrt(int n) {
int i=1;
while (i*ireturn i-(i*i!=n);
}
int main() {
constexpr int x = isqrt(1764); 
u
printf("%d", x);
}
Listing 1: A program illustrating 
constexpr
The 
isqrt
function computes the square root of the argument 
n
. Starting 
at 
1
, the function increments the local variable 
i
until 
i*i
is greater than or 
equal to 
n
. If 
i*i == n
, it returns 
i
; otherwise, it returns 
i-1
. Notice that the 
invocation of 
isqrt
has a literal value, so the compiler could theoretically 
compute the result for you. The result will only ever take on one value 
u
.
Compiling Listing 1 on GCC 8.3 targeting x86-64 with 
-O2
yields the 
assembly in Listing 2.
.LC0:
.string "%d"
main:
sub rsp, 8
mov esi, 42 
u
mov edi, OFFSET FLAT:.LC0
xor eax, eax
call printf
xor eax, eax
add rsp, 8
ret
Listing 2: The assembly produced after compiling Listing 1
The salient result here is the second instruction in 
main
u
; rather than 
evaluating the square root of 
1764
at runtime, the compiler evaluates it and 
outputs instructions to treat 
x
as 
42
. Of course, you could calculate the square 
root using a calculator and insert the result manually, but using 
constexpr
pro-
vides lots of benefits. This approach can mitigate many errors associated with 
manually copying and pasting, and it makes your code more expressive.

---

An Overture to C Programmers