4.8 Formulating Algorithms: Counter-Controlled Repetition

4.9 Formulating Algorithms: Sentinel-Controlled Repetition

4.10 Formulating Algorithms: Nested Control Statements

4.11 Assignment Operators

4.12 Increment and Decrement Operators

. . . . . . . . . ( end of section Prepare) <<Contents | End>>

Deliver a Question

When to use the different kinds of loop -- while for do-while

Everybody agrees that you must use the for loop for counting and scanning through data. The while is more general purpose and best at iterative algorithms.

I personally think you should never use the "do-while". Just an opinion based on 30 years experience.... I used them from 1964 through to 1978 and then abandoned them and started to have fewer bugs.

Pitfalls of the do-while

Basically having to write code in the do-while that works whether or not the condition is true.

Also changing a problem with zero-repetitions to one where there must be at least one repetition.

What is the purpose of Pseudocode

To be able to think about a solution of a problem without worrying about the C++ syntax.

What is the reason for nested control statements

Because we have nested problems, of course:-)

Seriously: In the lab I showed you a program that output a grid of symbols. The structure is (

1. row (
   1. item item item item ...
   )
2. row (
   1. item item item item ...
   )
3. ...
) so the only wise way to produce the output is a nested loop.

In general most probems demand a complicated algorithm to solve them... one where a selection has to be repeated, or where we have to choose between two loops. We can have solutions where a repeated pattern has to (itself) be repeated.

Is the member function determineClassAverage a C++ keyword

No. In fact, only keywords that are operators can be redefined as member functions. Don't do this in CS201!

Is the member function determineClassAverage assigned a value

No. Functions are not assigned values... you must define them. So we need to include a definition like this somewhere in our compilation:

 	void GradeBook::determineClassAverage()

 	{

 		//whatever

 	}

Should I use ++ and -- rather than writing the assignment out

Yes! Don't forget it is simpler to type

 		myLongNamedCounter++;

than

 		myLongNamedCounter=myLongNamedCounter+1;

Pre and post increment operators


Table

Before	Operation	After	Value returned
	v is v0	v++	v is v0+1	v0
	v is v0	++v	v is v0+1	v0+1
	v is v0	v+1	v is v0	v0+1
	v is v0	v--	v is v0-1	v0
	v is v0	--v	v is v0-1	v0-1
	v is v0	v-1	v is v0	v0-1


(Close Table)


The wise programmer uses "++" and "--" very simply to add and subtract 1 like his

 		v++;

and avoids complicated trickery like this:

 		v= (++v) * (v--);

which takes a lot of thought to get right and even more thought to figure out.

Do assignment operators help or just be confusing

Both. It depends on experience and the precise use. Many people use them and as a result you have to be able to read them.

Usually it does not make much difference. But when you have a long variable and want to add 2 to it:

	myObject.forExampleAttribute0+=2

it is too easy to mistype the expanded version:

	myObject.forExampleAttribute0= myObject.forExampleAttributeO+2

If output shows an object to have value zero does it mean it wasn't allocated a value

Not really, it is easy for a variable to become zero as the result of a complex calculation.

Also -- there is no guarantee that ints and doubles are set to zero initially.

Can you set a repetition to terminate at any value

Yes.

Is it better to use a merge symbol in UML for multiple activities

When the UML activity diagram or flowchart shows a choice of two or more alternatives... it should show the end of the selection with merge "---<>--" symbols.

What is the purpose of a sentinel symbol if it is not acceptable value

It's value is that it can not be used as input to the algorithm and so it can be used to signal the need to move on to something else.

How does float differ from double

Floats don't use as many binary digits, are more approximate, and cover a smaller range of real numbers.

Examples of conversions

[ convert.cpp ]

Explain implicit and explicit conversion

In C++ you can often convert data of one type to another type -- for example converting an integer into a double. If you use a "cast" then it is explicit.

In an implicit conversion, you allow the compiler to choose a conversion from one number type to another when necessary.

Notice that if you write

 		double answer = 1/2;

the division is int division and so answer is set to zero. However, if one of the divisors was a double

 		double answer = 1/2.0;

then we get answer set to 0.5. Neither does "double(1/2)" work. The division is done first (using int division) and then is converted to double. :! As a rule it is wise to use explicit conversions.

How are cast operators used

These force numbers from one form to another: double to float to int to char. The old fashioned notation was to use the target type. So to find the character whose code is 32 we write:

 		char(32)

(in ASCII a space character). Similarly to get an accurate average from the sum of a number of int's we used to write:

 		average = double(sum)/number;

This still works but is deprecated (jargon for: "will not work forever").

The more modern and complex form is shown on line 81 of the listing on page 158 of Deitel:

 		average = static_cast<double>(sum)/number;

this should continue to compile and run correctly even if the standard changes. And so this is the form you should use:-(

How can I convert numbers to strings and strings to numbers

I've dug up a couple of advanced types of objects "stringstreams" that do the job if you need it in a project, for example. Here [ stringnumbers.cpp ] is a program that demonstrates how to use a couple of functions "convertDouble" and "convertToDouble" that are defined in [ stringnumbers.h ] that you can download and #include in any of your programs.

Similar code can handle other numeric data and let you choose the formatting.

How do you output numbers to only 2 decimals

Use an IO Manipulator like this

 		cout << .... << setprecision(2) << ....;

[ ../examples/giomanip.cpp ] [ ../examples/iomanip.cpp ]

This changes the format of doubles for the whole prgram.... until you change it to some other precision.

Avoiding infinite loops

The only way to avoid infinite loops is to think. You must be sure, or better, must have proved that each time through the body, you are closer to the terminating condition. For example, if you are counting up, you condition should be counter < some_limit. If you are going down then the condition will be counter > some_value. Avoid using equality tests, by the way, in case the variable jumps over the limit. This also stops you type variable=limit when you want variable==limit.

The example in the book showed the need for braces {...} to include the commands that change the controlled variable in the body of the loop.

It is important that you understand the syntax of loops and use braces when needed. Indeed I tend to put them in in every loop, just in case.

But, there is no replacement for thought when coding loops and selections!

Why can uninitialized variables lead to errors

Because 'int's and 'double's are not set up to be any particular value. So if you are adding up data in an uninitialized variable the results are unpredictable, and probably wrong.

How do I make sure the user has given me one the small number of numbers I expect

This is a Data Validation problem. Let us assume that the user must input either 1, 2, or 3. Also assume that the user know enough to not input anything but an integer (not Roman "III", or a real number "1.234". Then we need to (
1. Input a number
2. if the number is 1 it is valid
3. else if the number is 2 it is valid
4. else if the number is 3 it is valid
5. else it is invalid and the user should try again.
)

This looks like a loop -- but how can we check the loop condition?

Here is a standard way to tackle it using the tools covered in the course so far.

 	int valid = 0; // 0 means invalid, 1 means valid

 	while(valid==0)

 	{	cout << "Input either 1, 2, or 3: ";

 		cin >> number;

 		if (number==1)

 			valid=1;

 		else if (number==2)

 			valid=1;

 		else if (number==3)

 			valid=1;

 		if(valid==0)

 			cout << number << " is invalid\n";

 	}

Using Boolean Expressions and variables from Chapter 5 makes this much easier!