For loop construction and code complexity [closed] - for-loop

Question

Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 4 years ago.
Improve this question
My group is having some discussion and strong feelings about for loop construction.
I have favored loops like:
size_t x;
for (x = 0; x < LIMIT; ++x) {
if (something) {
break;
}
...
}
// If we found what we're looking for, process it.
if (x < LIMIT) {
...
}
But others seem to prefer a Boolean flag like:
size_t x;
bool found = false;
for (x = 0; x < LIMIT && !found; ++x) {
if (something) {
found = true;
}
else {
...
}
}
// If we found what we're looking for, process it.
if (found) {
...
}
(And, where the language allows, using "for (int x = 0; ...".)
The first style has one less variable to keep track of and a simpler loop header. Albeit at the cost of "overloading" the loop control variable and (some would complain), the use of break.
The second style has clearly defined roles for the variables but a more complex loop condition and loop body (either an else, or a continue after found is set, or a "if (!found)" in the balance of the loop).
I think that the first style wins on code complexity. I'm looking for opinions from a broader audience. Pointers to actual research on which is easier to read and maintain would be even better. "It doesn't matter, take it out of your standard" is a fine answer, too.
OTOH, this may be the wrong question. I'm beginning to think that the right rule is "if you have to break out of a for, it's really a while."
bool found = false;
x = 0;
while (!found && x < LIMIT) {
if (something) {
found = true;
...handle the thing...
}
else {
...
}
++x;
}
Does what the first two examples do but in fewer lines. It does divide the initialization, test, and increment of x across three lines, though.

Answer 1

I'd actually dare to suggest consideration of GOTO to break out of loops in such cases:
for (size_t x = 0; x < LIMIT && !found; ++x) {
if (something)
goto found;
else {
...
}
}
// not found
...
return;
found:
...
return;
I consider this form to be both succint and readable. It may do some good in many simple cases (say, when there is no common processing in this function, in both found/unfound cases).
And about the general frowning goto receives, I find it to be a common misinterpretation of Dijkstra's original claims: his arguments favoured structured loop clauses, as for or while, over a primitive loop-via-goto, that still had a lot of presence circa 1968. Even the almighty Knuth eventualy says -
The new morality that I propose may
perhaps be stated thus: "Certain go to
statements which arise in connection with
well-understood transformations are acceptable, provided that the program documentation explains what the transformation was."
Others here occasionaly think the same.

Answer 2

While I disagree that an extra else really makes the 2nd more complicated, I think it's primarily a matter of aesthetics and keeping to your standard.
Personally, I have a probably irrational dislike of breaks and continues, so I'm MUCH more likely to use the found variable.
Also, note that you CAN add the found variable to the 1st implementation and do
if(something)
{
found = true;
break;
}
if you want to avoid the variable overloading problem at the expense of the extra variable, but still want the simple loop terminator...

Answer 3

The former example duplicates the x < LIMIT condition, whereas the latter doesn't.
With the former, if you want to change that condition, you have to remember to do it in two places.

Answer 4

I would prefer a different one altogether:
for (int x = 0; x < LIMIT; ++x) {
if (something) {
// If we found what we're looking for, process it.
...
break;
}
...
}
It seems you have not any trouble you mention about one or the other... ;-)
no duplication of condition, or readability problem
no additional variable

Answer 5

I don't have any references to hand (-1! -1!), but I seem to recall that having multiple exit points (from a function, from a loop) has been shown to cause issues with maintainability (I used to know someone who wrote code for the UK military and it was Verboten to do so). But more importantly, as RichieHindle points out, having a duplicate condition is a Bad Thing, it cries out for introducing bugs by changing one and not the other.
If you weren't using the condition later, I wouldn't be bothered either way. Since you are, the second is the way to go.

Answer 6

This sort of argument has been fought out here before (probably many times) such as in this question.
There are those that will argue that purity of code is all-important and they'll complain bitterly that your first option doesn't have identical post-conditions for all cases.
What I would answer is "Twaddle!". I'm a pragmatist, not a purist. I'm as against too much spaghetti code as much as the next engineer but some of the hideous terminating conditions I've seen in for loops are far worse than using a couple of breaks within your loop.
I will always go for readability of code over "purity" simply because I have to maintain it.

Answer 7

This looks like a place for a while loop. For loops are Syntactic Sugar on top of a While loop anyway. The general rule is that if you have to break out of a For loop, then use a While loop instead.

Answer 8

package com.company;
import java.io.*;
import java.util.Scanner;
public class Main {
// "line.separator" is a system property that is a platform independent and it is one way
// of getting a newline from your environment.
private static String NEWLINE = System.getProperty("line.separator");
public static void main(String[] args) {
// write your code here
boolean itsdone = false;
String userInputFileName;
String FirstName = null;
String LastName = null;
String user_junk;
String userOutputFileName;
String outString;
int Age = -1;
int rint = 0;
int myMAX = 100;
int MyArr2[] = new int[myMAX];
int itemCount = 0;
double average = 0;
double total = 0;
boolean ageDone = false;
Scanner inScan = new Scanner(System.in);
System.out.println("Enter First Name");
FirstName = inScan.next();
System.out.println("Enter Last Name");
LastName = inScan.next();
ageDone = false;
while (!ageDone) {
System.out.println("Enter Your Age");
if (inScan.hasNextInt()) {
Age = inScan.nextInt();
System.out.println(FirstName + " " + LastName + " " + "is " + Age + " Years old");
ageDone = true;
} else {
System.out.println("Your Age Needs to Have an Integer Value... Enter an Integer Value");
user_junk = inScan.next();
ageDone = false;
}
}
try {
File outputFile = new File("firstOutFile.txt");
if (outputFile.createNewFile()){
System.out.println("firstOutFile.txt was created"); // if file was created
}
else {
System.out.println("firstOutFile.txt existed and is being overwritten."); // if file had already existed
}
// --------------------------------
// If the file creation of access permissions to write into it
// are incorrect the program throws an exception
//
if ((outputFile.isFile()|| outputFile.canWrite())){
BufferedWriter fileOut = new BufferedWriter(new FileWriter(outputFile));
fileOut.write("==================================================================");
fileOut.write(NEWLINE + NEWLINE +" You Information is..." + NEWLINE + NEWLINE);
fileOut.write(NEWLINE + FirstName + " " + LastName + " " + Age + NEWLINE);
fileOut.write("==================================================================");
fileOut.close();
}
else {
throw new IOException();
}
} // end of try
catch (IOException e) { // in case for some reason the output file could not be created
System.err.format("IOException: %s%n", e);
e.printStackTrace();
}
} // end main method
}