MESSAGE
| DATE | 2010-02-14 |
| FROM | Ruben Safir
|
| SUBJECT | Re: [NYLXS - HANGOUT] C++ Workshop _ Syntax Basics
|
Data Assignments in C++ Statements:
We've now bootstrapped enough background information in order to examine
data assignment and variables in C++ statements, and to compile small
programs to explore C++ data better.
C and C++ are know as compiled typed languages. What this means is that
the C source code itself does not directly run a program, unlike
scripting languages like Perl, Python, Rudy, Bourn Shell Scripting and
such. The program has to be compiled, linked with standard library
binary libraries, and outputted into a binary file that the machine and
operating system can run.
The TYPED aspect of C++ means that all variables have to be defined as
some specific data type, one of the data types such as we discussed
earlier (such as int, char, short, long, double, float). Functions and
Methods are data and have a type. Its a code type. But for the
purposes of Syntax they are typed according to the kind of data that
they return.
When we create variables in C and C++ they have to be typed. In
addition, there is 3 phases in variable creation, which can be either in
separate statements of combined into one statement. The three phases
are:
Declaration
Definition (needed for functions and methods)
Initialization
We declare a variable using the name and its type, its symbol and is
initializing it, we include an assignment of data. For example:
int i;
declares a variable i which is of type int, and most usually defines it.
float j;
declares a variable j of type float, as it is defined specifically to
your architecture and environment. We can then assign an integer to i
and a float to j using the following syntax.
#include
using namespace std;
int i;
float j; //Declarations of variables
int main(int argv, char * argc[])
{
i = 10;
j = 5.6;
return 0;
}
Now here is the catch. WHEN WE DECLARE A VARIABLE in an ordinary
C++ coding file, one that has the function main within it, placing the
variable in the global namespace above or outside of the "main"
function, then we are defining and declaring the variable and the
variable exists globally for your program, from the point that the
variable is declare on to the end of the file, and all the other
functions and code that you bring into your program from that point
forward. However, it is normal for real programming for one to use at
least three files for your program. Your main C++ code file, your
library source file which is used to create reusable objects in your
code base, and then finally, in HEADER files, the .h files that are
imported with
#ifndef MYOWNLIB
#include "myownlibrary"
#define MYOWNLIB
#endif
statements.
However, it is a programming error to ***define*** variables within your
headers. It can confuse the linker, because then they compiler will
make space for all the defined variables in every object that uses the
header file in your program, and all those objects may not agree as
to their values later on. So you want to declare it, so that everyone
knows what the common type and symbol is, but you don't want to define
it, so that the linker doesn't come along and fill in all those
definitions, possibly incorrectly, and separately if not independently,
in all the objects of your complex program. In order to allow for the
declaration without the definition of a variable in the header file, you
must use the "extern" keyword in header files to create proper global
scope of variables.
This is not true with functions or classes, which we will see later.
They have implicit "extern"s.
To beat a dead horse, since this is one of the most common areas of
programming bugs, understand this:
Where we declare a variable is important and determines where your
variable is viewable in your program. Complex programs have so many
variables in them that it is critical to try to restrict their access
and usage to the smallest subset of need as practical. This is
accomplished through two mechanisms, Scope and Namespace. We've seen
namespace already with the using directive
using namespace std;
imports into our section of programming all the reserved symbols of the
standard namespace. Scope, on the other hand, restricts variable access
to specific blocks of our programs, not just the symbols. When we
declare our variables outside of main, as we did in the program
file1.cc, then the scope is considered to be global and the variables i
and j are available throughout our program.
Scope is a bit more complex We've looked at global scope already and we
will run up against it time and again. According to Lippman there are
three
kinds of Scope: local, namespace and class. Scope in C++ is fairly
complicated, and not well described in the standard texts. Anywhere you
have a block you can declare and define variables, and they are local to
that block. This is particularly important in functions. A block is a
group of statements which are surrounded by curly braces. Like the unix
shell, a block inherits the variables in its scope from the name space
it is called from, including blocks that wrap it. So you can have
blocks within blocks. If you declare a variable already in play, it
creates a new temporary version of that variable. Variables in
functions, while they might be declared outside the function, and
defined within a function, would be created by the compiler, but don't
become active until they are called. Each function creates a stack of
runtime processes. The top most process, ignoring threading and forked
processes for the moment, needs to run its course and end prior to
calling function can continue. Since everything begins with a function
called main, a runtime stack might look like this, each platter on the
stack having its own unique environment and variables scoped within,
excluding global variables which they can affect and see.
function2
|
function1-------|
|
main ---|
In this scenario, when function2 finishes, all of its local variables
die and no long exist in runtime, and control returns back to function1.
function1 can then call function3, and add new plates to the runtime
stack. We will explore this more when we look at functions and methods.
Now lets return to the representation of data in C++ and look at
programming examples.
I will now increasingly be showing example code on the NYLXS website
because of nice HTML tool in VIM that can show code in syntax color.
See http://www.nylxs.com/docs/workshops/?C=M;O=D for coding examples as
mentions here.
We can start by building a nice example program, which will have 4
working files, a header file, a main programming file, a library
programming file and a make file. There are easier ways of showing
these data type examples, but I feel it is important that new C and C++
programmer become comfortable and familiar with multiple file
programming projects, especially for C++. Also, I might mention a tip
or two with VIM and GVIM, in order to outline more general GNU tool sets
to help the C and C++ coder. There are also, no doubt, likewise
examples in EMACS and other editors, but I'm not going to mention them,
and I really don't want an editor war to break out in the workshop.
|
|
