# Is it better to put more logic in a for loop condition, or to use a while loop?

I had this as an interview question, and the interviewer pointed this out. Here's what I wrote:

//C# Syntax here
public string Reverse(string s)
{
char[] arr = s.ToCharArray();
int idx = 0;
int endIdx = s.Length - 1;
for(; idx < endIdx; ++idx, --endIdx)
{
char temp = arr[idx];
arr[idx] = arr[endIdx];
arr[endIdx] = temp;
}
return arr.ToString();
}


However, the interviewer was asking if I should have changed it to this:

//C# Syntax here
public string Reverse(string s)
{
char[] arr = s.ToCharArray();
int idx = 0;
int endIdx = s.Length - 1;
while(idx < endIdx)
{
char temp = arr[idx];
arr[idx] = arr[endIdx];
arr[endIdx] = temp;
++idx;
--endIdx;
}
return arr.ToString();
}


Personally, I like the first version, because it puts all of the machinery which controls the loop into the loop statement. One can insert continue or break or return statements into the loop without it breaking things.

However, some programmers dislike putting that stuff into the for, because they think it's being too "clever".

What do you consider best-practice for this?

-

## migrated from programmers.stackexchange.comMar 31 '11 at 16:26

This question came from our site for professional programmers interested in conceptual questions about software development.

I think this would be a better question for codereview.stackexchange.com or stackoverflow. –  whatsisname Mar 31 '11 at 16:15
@whatisname: Maybe code review. I don't think it works on SO because it's subjective. –  Billy ONeal Mar 31 '11 at 16:18
@Brian: I disagree with that tag -- this particular example is using C# as the language, but I'm asking about any language where for exists in the form taken from C. –  Billy ONeal Apr 1 '11 at 2:49
IMHO you was posted this correctly to Programmers. ;p As you noted the example code isn't really meant to be reviewed, but just an example. –  Steven Jeuris Apr 11 '11 at 10:49

I would argue against changing it. This seems to be exactly what for-loops were invented for. Also, why not put the initialization in the for loop?

I would have also preferred

for(int idx=0; idx < endIdx; ++idx, --endIdx) {


if only so that I don't have to scan up and down the page looking for the initialization of idx.

-
Because this was on a whiteboard, and I was running out of space :P (And because that makes the FOR line long) –  Billy ONeal Mar 31 '11 at 16:19
@Billy ONeal: Ok, whiteboard space constraints are fair ;) –  FrustratedWithFormsDesigner Mar 31 '11 at 16:21
Why not: for (int idx = 0, endIdx = s.Length-1; idx < endIdx; ++idx, --endIdx) { ... }? That way you get all the loop controls in the loop. –  Jonathan Leffler Apr 3 '11 at 4:10
@Billy ONeal: I prefer having all loop counters/conditionals inside the for-loop, and if that makes the for-loop line too long, I'd break them into three lines. –  Lie Ryan Apr 11 '11 at 11:08

I personally would go with the while loop. I'm not against for loops at all, but yours is different than most, and caused me to reread to catch up. I am used to seeing, as I imagine most people are, for loops with an initializer and only one statement to iterate:

for (int i = startValue; i < endValue; i++) {...}

To me, you're already initializing outside of the loop statement (as you said, you prefer to keep things in the loop statement), so you may as well use a while so people expect initialization to be elsewhere. Same with using this construct:

++idx, --endIdx

I'm not expecting this sort of thing in a for loop. I would be more aware something like this might happen in a while loop.

Summary: I personally think the while loop is more readable, because I have certain expectations for for loops. If this is true of the other people who will read/maintain the code, I don't know.

-
Do you have rationale other than "this is the one I like"? –  Billy ONeal Mar 31 '11 at 20:06
I feel like it adheres better to user expectations, where "users" in this case are other people who will work with the code. Now yes, I base that on what I personally expect, as detailed in my answer. The for loop you have was more difficult for me to read, because it did things I don't normally see in for loops, but more frequently see in while loops. So yes, it's pretty subjective, but not just because I fancy while loops or have a thing against the word "for" or something :) –  PeterL Mar 31 '11 at 21:19
Okay, that's reasonable. +1. –  Billy ONeal Apr 1 '11 at 2:51

That code looks pretty complex for the trivial task it does.

1. No need to convert the string to a char array. By not calling ToCharArray() you get rid of an extra copy of data.
2. Poorly named variables IMHO. (The convention for index is i.)
3. No need to substract from the end index. (which I prefer to call length as a convention)
4. You shouldn't call ToString(), but use the constructor of string. Your code outputs the typename instead of the reversed string.

Altogether:

public string Reverse( string s )
{
int length = s.Length;
char[] reversed = new char[ length ];

for ( int i = 0; i < length; ++i )
{
reversed[ i ] = s[ length - 1 - i ];
}

return new string( reversed );
}


And of course, the proper solution would be using Reverse() from IEnumerable. But that wasn't part of the question perhaps.

return new string( s.Reverse().ToArray() );


An interesting article which compares different reverse string implementations can be found here.

It's a pity you got referred to here from Programmers.SE, IMHO you do touch some points which are perfectly discussable on Programmers.

-
1. Yes there is. Just because you did it a different way doesn't mean you didn't do it. 2. I strongly disagree. Single letter variable names make me cringe. The counter is primarily an array index, and it's name reflects that. 3. Okay, I agree that's nicer. 4. Why does it matter? 5. READ THE QUESTION. I'm not talking about this specific implementation of Reverse. I'm talking about use of logic in for loops. Reverse is the algorithm I happened to be using as an example when the question came up, but critiquing how I threw this together in 30 seconds is not the point. –  Billy ONeal Apr 8 '11 at 4:58
1. By not calling ToCharArray() you get rid of an extra copy of data. 2. The name of a variable should be only as important as the scope it is in. The for loop indexer is a perfect example of a small scope. If you don't like i, I'd still prefer index over idx, but I know of practically noone who doesn't use i. 4. Your code outputs the typename instead of the reversed string. 5. When you post on Code Review, expect a code review. If you want to discuss a general topic go to Programmers. –  Steven Jeuris Apr 8 '11 at 5:25
I didn't post to code review. I posted to programmers. It got moved here. (Sorry for shouting last night -- I was half asleep and not thinking straight.) –  Billy ONeal Apr 8 '11 at 18:07
Sorry @rolfl but this answer is not wrong. Once an index is set in the reversed array it is not modified again. The for-loop reads from s and writes to reversed. –  Simon André Forsberg Dec 10 '13 at 13:11
@rolfl I don't swap anything (as in the OPs solution). I iterate across all chars in reverse order, for simplicity reasons. Performance wise one might be better than the other, but I'd need to run into a performance issue prior to refactoring that. –  Steven Jeuris Dec 10 '13 at 13:12

The primary consideration when deciding what to put in the for-loop header should be whether the for-loop header conveys the structure of the loop. The three elements (initializer; condition; update) should form a coherent story.

I would express your function like this:

public string Reverse(string s)
{
char[] arr = s.ToCharArray();
for (int i = 0, j = s.Length - 1; i < j; ++i, --j)
{
char swap = arr[i];
arr[i] = arr[j];
arr[j] = swap;
}
return arr.ToString();
}


From just the for-loop header, a reasonably experienced programmer can recognize the idiom for stepping two array indices until they meet in the middle. By convention, i and j are array indices; you don't need the …Idx Hungarian suffix.

In comparison, the interviewer's proposal looks unstructured. I feel like I have to mentally decompile it back into the for-loop above to understand what is going on.

I've also renamed temp to swap as a silent "comment".

You might be wondering, to what extreme can you pack logic into the for-loop header? In practice, I've never seen more a good for-loop header that involved more than two variables. The probable explanation is that the condition usually involves a binary comparison operator. A third variable would therefore usually be "off-topic" for the header.

Two other guidelines you might want to use are:

• If you can fill in all three parts of the header (initializer; condition; update), then a for-loop is probably appropriate.
• If the loop body surreptitiously updates the iteration variable, such that the header tells a misleading or incomplete story, then the iteration should probably not be done using a for-loop.
-

Personally, I can't really warm up to either version. I guess that to me, reversing a sequence is so obviously a fold, that anything which doesn't implement it as a fold looks just plain weird.

I would have expected to see something much more like this:

public static string Reverse(this string s)
{
return s.Aggregate("", (acc, el) => el + acc);
}


I.e. implemented as a left fold.

After all, reversing any sequence (not just strings) is really just the same as

foldl (flip (:)) []

-
That performs horribly though (in the first case -- you've turned a constant time algorithm into an O(N^2) one, with a memory allocation for the whole string for each character!)). I don't know what fold1 is. (Nor have I ever heard of the concept) –  Billy ONeal Apr 1 '11 at 2:40
More to the point, this doesn't answer my question -- I'm not asking about this specific implementation of reverse, I'm asking about the general case of embedding logic more complicated than "n + 1" in a for loop condition. –  Billy ONeal Apr 1 '11 at 2:47
Err.. where I said "constant time" above I meant "linear time". –  Billy ONeal Apr 1 '11 at 2:48
In a functional language, a fold or more explicit recursion, would often replace a loop. Single-Linked-Lists are also very common in functional languages, where a string would be a linked-list of char. And actually, a properly done functional algorithm would still be O(n), since you'd go over the list, building a new list in reverse order as you went. –  CodexArcanum Apr 1 '11 at 22:12
Such as: reverse first :: rest = (reverse rest) :: first And pretend I added the case for empty list. :: is the "cons" operator. Check out F# for functional programming in .NET. –  CodexArcanum Apr 1 '11 at 22:14

I honestly don't think it matters. For a function this simple, I can think of several similar implementations:

• Have an index to the front and to the back in a for loop
• Have an index to the front and to the back in a while loop
• Have a single index, and calculate the offset from the end of the loop each time round the loop
• All of the possibilities above, but increment/decrement a pointer to char, rather than an index into an array (at least in C, can you do this in C# ?)

And probably several more. I think that the right answer is whatever seems most readable (unless it's EGREGIOUSLY inefficient), and if you think it may be called in an inner loop leave a comment saying (a) don't do that and (b) suggest what changes you'd think about making then, but you'd most likely need to be sure, the function is too simple for your common sense to be much guide about what's faster.

I agree your answer has the potential problems the interviewer described, and his has the potential problems you described. I think either is more than fine, so long as you can see the potential issues, I don't think they can be improved into a single "best" answer.

-
"can you do this in C#" <-- Not really. It's possible but you'll have to go into "unsafe code" -- which means your code won't run several places, i.e. Silverlight, because unsafe code requires full trust. Again though, I'm not talking about this specific case of reverse. I'm talking about the general case -- this is just the algorithm I happened to be talking about when the general question came up. –  Billy ONeal Apr 1 '11 at 16:49

I was going to write this as a comment but got out of space, so...

In general lines, I agree with Peter Leppert; I think the while is more idiomatic here.

Of course C for syntax allows you to use the first case, and in simple functions I don't think it really matters, but then, C allows you implementing mostly any function in the header of a for loop with an empty body, and that doesn't makes it a good idea ;-).

for(; idx < endIdx; ++idx, --endIdx)


just because you got out of space in the whiteboard. In a computer it would have been more like the following:

for(int idx = 0, endIdx = s.Length - 1; idx < endIdx; ++idx, --endIdx)


Ok, fair enough. Now you find you need another control variable for a new functionality:

for(int idx = 0, endIdx = s.Length - 1, otherVar = FunctionCall(param1, param2); idx < endIdx && otherVar < AnotherFunctionCall(param1); ++idx, --endIdx, otherVar += 5)


What now, uh? It gets ugly very quickly. This doesn't scale up, so, where do you put the line? Character limit in header? Only two control variables?

That's why I would only do the classic for loop:

for (int i = 0; i < maxI; i++)
{
...
}


and use a while for everything else.

Anyway, with that question the interviewer was probably trying to see how you think rather than testing your programming capabilities.

-
But if you limit the for loop to that only, then there's little reason for it to exist at all -- foreach handles that. (i.e. foreach(var idx = Enumerable.Range(0, maxI)) { ... }) (That syntax isn't 100% correct but it gets the idea) –  Billy ONeal Apr 3 '11 at 19:00
@Billy Oneal: Sorry, you're right, I was thinking in C++ where we still don't have a for each, so I didn't use the best example. However, it still applies if you don't want to process every item in the list, or need to do so in specific order (i.e. traverse a list backwards) –  Jaime Pardos Apr 10 '11 at 20:45
@BillyONeal, @JaimePardos is probably correct that the interviewer just wanted to see your thought process on tackling this problem. IMHO, I like the for for the simple reason that it's easier to read and you don't have to worry about setting the start and end index for each loop. –  Dennis Rongo Dec 9 '13 at 20:31

Although both for and while are loops and both will work but as a general guideline, for loop is better suited in this case because

• When we know the number of iterations in advance, we should use for loop as it allows initiation, condition and increment to be done in one line.
• Index variable (idx in this case), is more appropiately scoped, why should we increase the scope of idx when this is meant only to be used within the loop.
• Unlike while loop, statements like continue, etc. (if required) can be used without any issues.

While loop should be usually used if we don't know number of iterations in advance.

-

I think it's important to go back to the context that this was an interview question.

When I interview people, I try to ask questions that don't have one "right" answer - that's more like a trivia test, which is actually fairly useless to me. Instead, I'll ask a question that requires more of an opinion response, watch the candidate, and look for them to apply logical thinking to the problem.

If you can justify that it should be a for loop, and your code is readable, good answer. If you can justify that it should be a while loop, and your code is readable, that's also a good answer. But if you say "it should be a for loop because that's the right way," that's just spewing out dogma and doesn't demonstrate independent thought. It's a less mature answer. Not that it's wrong, but if I'm trying to figure out if you're a starting programmer or an experienced engineer, it's a clue.

By the way, another really good answer is to say "In my code I'd use Reverse() on the array, because it's already tested." An employer generally does not want to pay you to reinvent the wheel, and as an interviewer I'd appreciate the depth of your experience that shows.

-

@Billy ONeal : I strongly believe that this is a matter of taste. Robert C. Martin states in Clean Code that software developers like to use obfuscated code to "show off" and I agree with him. I also liked to do that until I started to work on my first complex project and I realized that extra mind mappings are not necessary.

That been said, my option will be to use the while() loop; I think that we can go here with the same principle which is applied to a method: the loop should do one thing and one thing only (personal opinion extrapolated from Clean Code). So, if I look at a piece of code after 4 hours of endless debugging, it might feel easier to read (I only formatted the code for easier reading):

     //top-down reading per loop
while (UnswitchedCharactersExist())
{
SwitchCharacters(arr, idx, endIdx);
IncrementStartIndex();
DecrementEndIndex();
}


than:

     //read the initializer, the condition, the code block and go back up for the update
//after that read the condition, the code block and go back up for the update
for(; UnswitchedCharactersExist(); IncrementStartIndex(), DecrementEndIndex())
{
SwitchCharacters(arr, idx, endIdx);
}


My 2 cents on the subject.

-

All things aside, I believe that these two pieces of code function differently. The first is predecrement, the second is post.

I would agree that for loops make this specific application more obfuscated (not by much, but it is certainly less clear). I recommend using for loops primarily for iterating linearly through a collection with the intent of going from some start to an end, where both of these points are known/fixed before the loop starts.

I think that the for loop is a completely valid approach for something like this though. If you find yourself making a for loop and placing an if statement nearly in the beginning of it, then you may want to consider using a while loop.

-
Erm, no, both cases are preincrement. –  Billy ONeal Dec 10 '13 at 4:51

## protected by rolfl♦Dec 10 '13 at 1:40

Thank you for your interest in this question. Because it has attracted low-quality answers, posting an answer now requires 10 reputation on this site.