If I have a for loop which is nested within another, how can I efficiently come out of both loops (inner and outer) in the quickest possible way?
I don't want to have to use a boolean and then have to say go to another method, but rather just to execute the first line of code after the outer loop.
What is a quick and nice way of going about this?
I was thinking that exceptions aren't cheap/should only be thrown in a truly exceptional condition etc. Hence I don't think this solution would be good from a performance perspective.
I don't feel it it is right to take advantage of the newer features in .NET (anon methods) to do something which is pretty fundamental.
This question is related to
c#
for-loop
nested-loops
Use a suitable guard in the outer loop. Set the guard in the inner loop before you break.
bool exitedInner = false;
for (int i = 0; i < N && !exitedInner; ++i) {
.... some outer loop stuff
for (int j = 0; j < M; ++j) {
if (sometest) {
exitedInner = true;
break;
}
}
if (!exitedInner) {
... more outer loop stuff
}
}
Or better yet, abstract the inner loop into a method and exit the outer loop when it returns false.
for (int i = 0; i < N; ++i) {
.... some outer loop stuff
if (!doInner(i, N, M)) {
break;
}
... more outer loop stuff
}
Sometimes nice to abstract the code into it's own function and than use an early return - early returns are evil though : )
public void GetIndexOf(Transform transform, out int outX, out int outY)
{
outX = -1;
outY = -1;
for (int x = 0; x < Columns.Length; x++)
{
var column = Columns[x];
for (int y = 0; y < column.Transforms.Length; y++)
{
if(column.Transforms[y] == transform)
{
outX = x;
outY = y;
return;
}
}
}
}
Don't quote me on this, but you could use goto as suggested in the MSDN. There are other solutions, as including a flag that is checked in each iteration of both loops. Finally you could use an exception as a really heavyweight solution to your problem.
GOTO:
for ( int i = 0; i < 10; ++i ) {
for ( int j = 0; j < 10; ++j ) {
// code
if ( break_condition ) goto End;
// more code
}
}
End: ;
Condition:
bool exit = false;
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
exit = true;
break; // or continue
}
// more code
}
}
Exception:
try {
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
throw new Exception()
}
// more code
}
}
catch ( Exception e ) {}
The cleanest, shortest, and most reusable way is a self invoked anonymous function:
One line shorter than the top answer with anonymous method.
new Action(() =>
{
for (int x = 0; x < 100; x++)
{
for (int y = 0; y < 100; y++)
{
return; // exits self invoked lambda expression
}
}
})();
Console.WriteLine("Hi");
I remember from my student days that it was said it's mathematically provable that you can do anything in code without a goto (i.e. there is no situation where goto is the only answer). So, I never use goto's (just my personal preference, not suggesting that i'm right or wrong)
Anyways, to break out of nested loops I do something like this:
var isDone = false;
for (var x in collectionX) {
for (var y in collectionY) {
for (var z in collectionZ) {
if (conditionMet) {
// some code
isDone = true;
}
if (isDone)
break;
}
if (isDone)
break;
}
if (isDone)
break;
}
... i hope that helps for those who like me are anti-goto "fanboys" :)
C# adaptation of approach often used in C - set value of outer loop's variable outside of loop conditions (i.e. for loop using int variable INT_MAX -1 is often good choice):
for (int i = 0; i < 100; i++)
{
for (int j = 0; j < 100; j++)
{
if (exit_condition)
{
// cause the outer loop to break:
// use i = INT_MAX - 1; otherwise i++ == INT_MIN < 100 and loop will continue
i = int.MaxValue - 1;
Console.WriteLine("Hi");
// break the inner loop
break;
}
}
// if you have code in outer loop it will execute after break from inner loop
}
As note in code says break will not magically jump to next iteration of the outer loop - so if you have code outside of inner loop this approach requires more checks. Consider other solutions in such case.
This approach works with for and while loops but does not work for foreach. In case of foreach you won't have code access to the hidden enumerator so you can't change it (and even if you could IEnumerator doesn't have some "MoveToEnd" method).
Acknowledgments to inlined comments' authors:
i = INT_MAX - 1 suggestion by Meta
for/foreach comment by ygoe.
Proper IntMax by jmbpiano
remark about code after inner loop by blizpasta
Don't quote me on this, but you could use goto as suggested in the MSDN. There are other solutions, as including a flag that is checked in each iteration of both loops. Finally you could use an exception as a really heavyweight solution to your problem.
GOTO:
for ( int i = 0; i < 10; ++i ) {
for ( int j = 0; j < 10; ++j ) {
// code
if ( break_condition ) goto End;
// more code
}
}
End: ;
Condition:
bool exit = false;
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
exit = true;
break; // or continue
}
// more code
}
}
Exception:
try {
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
throw new Exception()
}
// more code
}
}
catch ( Exception e ) {}
You asked for a combination of quick, nice, no use of a boolean, no use of goto, and C#. You've ruled out all possible ways of doing what you want.
The most quick and least ugly way is to use a goto.
Did you even look at the break keyword? O.o
This is just pseudo-code, but you should be able to see what I mean:
<?php
for(...) {
while(...) {
foreach(...) {
break 3;
}
}
}
If you think about break being a function like break(), then it's parameter would be the number of loops to break out of. As we are in the third loop in the code here, we can break out of all three.
Manual: http://php.net/break
I've seen a lot of examples that use "break" but none that use "continue".
It still would require a flag of some sort in the inner loop:
while( some_condition )
{
// outer loop stuff
...
bool get_out = false;
for(...)
{
// inner loop stuff
...
get_out = true;
break;
}
if( get_out )
{
some_condition=false;
continue;
}
// more out loop stuff
...
}
Since I first saw break in C a couple of decades back, this problem has vexed me. I was hoping some language enhancement would have an extension to break which would work thus:
break; // our trusty friend, breaks out of current looping construct.
break 2; // breaks out of the current and it's parent looping construct.
break 3; // breaks out of 3 looping constructs.
break all; // totally decimates any looping constructs in force.
This solution does not apply to C#
For people who found this question via other languages, Javascript, Java, and D allows labeled breaks and continues:
outer: while(fn1())
{
while(fn2())
{
if(fn3()) continue outer;
if(fn4()) break outer;
}
}
It seems to me like people dislike a goto statement a lot, so I felt the need to straighten this out a bit.
I believe the 'emotions' people have about goto eventually boil down to understanding of code and (misconceptions) about possible performance implications. Before answering the question, I will therefore first go into some of the details on how it's compiled.
As we all know, C# is compiled to IL, which is then compiled to assembler using an SSA compiler. I'll give a bit of insights into how this all works, and then try to answer the question itself.
From C# to IL
First we need a piece of C# code. Let's start simple:
foreach (var item in array)
{
// ...
break;
// ...
}
I'll do this step by step to give you a good idea of what happens under the hood.
First translation: from foreach to the equivalent for loop (Note: I'm using an array here, because I don't want to get into details of IDisposable -- in which case I'd also have to use an IEnumerable):
for (int i=0; i<array.Length; ++i)
{
var item = array[i];
// ...
break;
// ...
}
Second translation: the for and break is translated into an easier equivalent:
int i=0;
while (i < array.Length)
{
var item = array[i];
// ...
break;
// ...
++i;
}
And third translation (this is the equivalent of the IL code): we change break and while into a branch:
int i=0; // for initialization
startLoop:
if (i >= array.Length) // for condition
{
goto exitLoop;
}
var item = array[i];
// ...
goto exitLoop; // break
// ...
++i; // for post-expression
goto startLoop;
While the compiler does these things in a single step, it gives you insight into the process. The IL code that evolves from the C# program is the literal translation of the last C# code. You can see for yourself here: https://dotnetfiddle.net/QaiLRz (click 'view IL')
Now, one thing you have observed here is that during the process, the code becomes more complex. The easiest way to observe this is by the fact that we needed more and more code to ackomplish the same thing. You might also argue that foreach, for, while and break are actually short-hands for goto, which is partly true.
From IL to Assembler
The .NET JIT compiler is an SSA compiler. I won't go into all the details of SSA form here and how to create an optimizing compiler, it's just too much, but can give a basic understanding about what will happen. For a deeper understanding, it's best to start reading up on optimizing compilers (I do like this book for a brief introduction: http://ssabook.gforge.inria.fr/latest/book.pdf ) and LLVM (llvm.org).
Every optimizing compiler relies on the fact that code is easy and follows predictable patterns. In the case of FOR loops, we use graph theory to analyze branches, and then optimize things like cycli in our branches (e.g. branches backwards).
However, we now have forward branches to implement our loops. As you might have guessed, this is actually one of the first steps the JIT is going to fix, like this:
int i=0; // for initialization
if (i >= array.Length) // for condition
{
goto endOfLoop;
}
startLoop:
var item = array[i];
// ...
goto endOfLoop; // break
// ...
++i; // for post-expression
if (i >= array.Length) // for condition
{
goto startLoop;
}
endOfLoop:
// ...
As you can see, we now have a backward branch, which is our little loop. The only thing that's still nasty here is the branch that we ended up with due to our break statement. In some cases, we can move this in the same way, but in others it's there to stay.
So why does the compiler do this? Well, if we can unroll the loop, we might be able to vectorize it. We might even be able to proof that there's just constants being added, which means our whole loop could vanish into thin air. To summarize: by making the patterns predictable (by making the branches predictable), we can proof that certain conditions hold in our loop, which means we can do magic during the JIT optimization.
However, branches tend to break those nice predictable patterns, which is something optimizers therefore kind-a dislike. Break, continue, goto - they all intend to break these predictable patterns- and are therefore not really 'nice'.
You should also realize at this point that a simple foreach is more predictable then a bunch of goto statements that go all over the place. In terms of (1) readability and (2) from an optimizer perspective, it's both the better solution.
Another thing worth mentioning is that it's very relevant for optimizing compilers to assign registers to variables (a process called register allocation). As you might know, there's only a finite number of registers in your CPU and they are by far the fastest pieces of memory in your hardware. Variables used in code that's in the inner-most loop, are more likely to get a register assigned, while variables outside of your loop are less important (because this code is probably hit less).
Help, too much complexity... what should I do?
The bottom line is that you should always use the language constructs you have at your disposal, which will usually (implictly) build predictable patterns for your compiler. Try to avoid strange branches if possible (specifically: break, continue, goto or a return in the middle of nothing).
The good news here is that these predictable patterns are both easy to read (for humans) and easy to spot (for compilers).
One of those patterns is called SESE, which stands for Single Entry Single Exit.
And now we get to the real question.
Imagine that you have something like this:
// a is a variable.
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a)
{
// break everything
}
}
}
The easiest way to make this a predictable pattern is to simply eliminate the if completely:
int i, j;
for (i=0; i<100 && i*j <= a; ++i)
{
for (j=0; j<100 && i*j <= a; ++j)
{
// ...
}
}
In other cases you can also split the method into 2 methods:
// Outer loop in method 1:
for (i=0; i<100 && processInner(i); ++i)
{
}
private bool processInner(int i)
{
int j;
for (j=0; j<100 && i*j <= a; ++j)
{
// ...
}
return i*j<=a;
}
Temporary variables? Good, bad or ugly?
You might even decide to return a boolean from within the loop (but I personally prefer the SESE form because that's how the compiler will see it and I think it's cleaner to read).
Some people think it's cleaner to use a temporary variable, and propose a solution like this:
bool more = true;
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a) { more = false; break; } // yuck.
// ...
}
if (!more) { break; } // yuck.
// ...
}
// ...
I personally am opposed to this approach. Look again on how the code is compiled. Now think about what this will do with these nice, predictable patterns. Get the picture?
Right, let me spell it out. What will happen is that:
more variable that only happens to be used in control flow. more will be eliminated from the program, and only branches remain. These branches will be optimized, so you will get only a single branch out of the inner loop.more is definitely used in the inner-most loop, so if the compiler won't optimize it away, it has a high chance to be allocated to a register (which eats up valuable register memory).So, to summarize: the optimizer in your compiler will go into a hell of a lot of trouble to figure out that more is only used for the control flow, and in the best case scenario will translate it to a single branch outside of the outer for loop.
In other words, the best case scenario is that it will end up with the equivalent of this:
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a) { goto exitLoop; } // perhaps add a comment
// ...
}
// ...
}
exitLoop:
// ...
My personal opinion on this is quite simple: if this is what we intended all along, let's make the world easier for both the compiler and readability, and write that right away.
tl;dr:
Bottom line:
goto or bool more, prefer the former.I've seen a lot of examples that use "break" but none that use "continue".
It still would require a flag of some sort in the inner loop:
while( some_condition )
{
// outer loop stuff
...
bool get_out = false;
for(...)
{
// inner loop stuff
...
get_out = true;
break;
}
if( get_out )
{
some_condition=false;
continue;
}
// more out loop stuff
...
}
C# adaptation of approach often used in C - set value of outer loop's variable outside of loop conditions (i.e. for loop using int variable INT_MAX -1 is often good choice):
for (int i = 0; i < 100; i++)
{
for (int j = 0; j < 100; j++)
{
if (exit_condition)
{
// cause the outer loop to break:
// use i = INT_MAX - 1; otherwise i++ == INT_MIN < 100 and loop will continue
i = int.MaxValue - 1;
Console.WriteLine("Hi");
// break the inner loop
break;
}
}
// if you have code in outer loop it will execute after break from inner loop
}
As note in code says break will not magically jump to next iteration of the outer loop - so if you have code outside of inner loop this approach requires more checks. Consider other solutions in such case.
This approach works with for and while loops but does not work for foreach. In case of foreach you won't have code access to the hidden enumerator so you can't change it (and even if you could IEnumerator doesn't have some "MoveToEnd" method).
Acknowledgments to inlined comments' authors:
i = INT_MAX - 1 suggestion by Meta
for/foreach comment by ygoe.
Proper IntMax by jmbpiano
remark about code after inner loop by blizpasta
You asked for a combination of quick, nice, no use of a boolean, no use of goto, and C#. You've ruled out all possible ways of doing what you want.
The most quick and least ugly way is to use a goto.
factor into a function/method and use early return, or rearrange your loops into a while-clause. goto/exceptions/whatever are certainly not appropriate here.
def do_until_equal():
foreach a:
foreach b:
if a==b: return
Throw a custom exception which goes out outter loop.
It works for for,foreach or while or any kind of loop and any language that uses try catch exception block
try
{
foreach (object o in list)
{
foreach (object another in otherList)
{
// ... some stuff here
if (condition)
{
throw new CustomExcpetion();
}
}
}
}
catch (CustomException)
{
// log
}
Is it possible to refactor the nested for loop into a private method? That way you could simply 'return' out of the method to exit the loop.
Since I first saw break in C a couple of decades back, this problem has vexed me. I was hoping some language enhancement would have an extension to break which would work thus:
break; // our trusty friend, breaks out of current looping construct.
break 2; // breaks out of the current and it's parent looping construct.
break 3; // breaks out of 3 looping constructs.
break all; // totally decimates any looping constructs in force.
Sometimes nice to abstract the code into it's own function and than use an early return - early returns are evil though : )
public void GetIndexOf(Transform transform, out int outX, out int outY)
{
outX = -1;
outY = -1;
for (int x = 0; x < Columns.Length; x++)
{
var column = Columns[x];
for (int y = 0; y < column.Transforms.Length; y++)
{
if(column.Transforms[y] == transform)
{
outX = x;
outY = y;
return;
}
}
}
}
This solution does not apply to C#
For people who found this question via other languages, Javascript, Java, and D allows labeled breaks and continues:
outer: while(fn1())
{
while(fn2())
{
if(fn3()) continue outer;
if(fn4()) break outer;
}
}
bool breakInnerLoop=false
for(int i=0;i<=10;i++)
{
for(int J=0;i<=10;i++)
{
if(i<=j)
{
breakInnerLoop=true;
break;
}
}
if(breakInnerLoop)
{
continue
}
}
You asked for a combination of quick, nice, no use of a boolean, no use of goto, and C#. You've ruled out all possible ways of doing what you want.
The most quick and least ugly way is to use a goto.
C# adaptation of approach often used in C - set value of outer loop's variable outside of loop conditions (i.e. for loop using int variable INT_MAX -1 is often good choice):
for (int i = 0; i < 100; i++)
{
for (int j = 0; j < 100; j++)
{
if (exit_condition)
{
// cause the outer loop to break:
// use i = INT_MAX - 1; otherwise i++ == INT_MIN < 100 and loop will continue
i = int.MaxValue - 1;
Console.WriteLine("Hi");
// break the inner loop
break;
}
}
// if you have code in outer loop it will execute after break from inner loop
}
As note in code says break will not magically jump to next iteration of the outer loop - so if you have code outside of inner loop this approach requires more checks. Consider other solutions in such case.
This approach works with for and while loops but does not work for foreach. In case of foreach you won't have code access to the hidden enumerator so you can't change it (and even if you could IEnumerator doesn't have some "MoveToEnd" method).
Acknowledgments to inlined comments' authors:
i = INT_MAX - 1 suggestion by Meta
for/foreach comment by ygoe.
Proper IntMax by jmbpiano
remark about code after inner loop by blizpasta
I think unless you want to do the "boolean thing" the only solution is actually to throw. Which you obviously shouldn't do..!
Use a suitable guard in the outer loop. Set the guard in the inner loop before you break.
bool exitedInner = false;
for (int i = 0; i < N && !exitedInner; ++i) {
.... some outer loop stuff
for (int j = 0; j < M; ++j) {
if (sometest) {
exitedInner = true;
break;
}
}
if (!exitedInner) {
... more outer loop stuff
}
}
Or better yet, abstract the inner loop into a method and exit the outer loop when it returns false.
for (int i = 0; i < N; ++i) {
.... some outer loop stuff
if (!doInner(i, N, M)) {
break;
}
... more outer loop stuff
}
This solution does not apply to C#
For people who found this question via other languages, Javascript, Java, and D allows labeled breaks and continues:
outer: while(fn1())
{
while(fn2())
{
if(fn3()) continue outer;
if(fn4()) break outer;
}
}
Don't quote me on this, but you could use goto as suggested in the MSDN. There are other solutions, as including a flag that is checked in each iteration of both loops. Finally you could use an exception as a really heavyweight solution to your problem.
GOTO:
for ( int i = 0; i < 10; ++i ) {
for ( int j = 0; j < 10; ++j ) {
// code
if ( break_condition ) goto End;
// more code
}
}
End: ;
Condition:
bool exit = false;
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
exit = true;
break; // or continue
}
// more code
}
}
Exception:
try {
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
throw new Exception()
}
// more code
}
}
catch ( Exception e ) {}
bool breakInnerLoop=false
for(int i=0;i<=10;i++)
{
for(int J=0;i<=10;i++)
{
if(i<=j)
{
breakInnerLoop=true;
break;
}
}
if(breakInnerLoop)
{
continue
}
}
I've seen a lot of examples that use "break" but none that use "continue".
It still would require a flag of some sort in the inner loop:
while( some_condition )
{
// outer loop stuff
...
bool get_out = false;
for(...)
{
// inner loop stuff
...
get_out = true;
break;
}
if( get_out )
{
some_condition=false;
continue;
}
// more out loop stuff
...
}
That's how I did it. Still a workaround.
foreach (var substring in substrings) {_x000D_
//To be used to break from 1st loop._x000D_
int breaker=1;_x000D_
foreach (char c in substring) {_x000D_
if (char.IsLetter(c)) {_x000D_
Console.WriteLine(line.IndexOf(c));_x000D_
\\setting condition to break from 1st loop._x000D_
breaker=9;_x000D_
break;_x000D_
}_x000D_
}_x000D_
if (breaker==9) {_x000D_
break;_x000D_
}_x000D_
}Since I first saw break in C a couple of decades back, this problem has vexed me. I was hoping some language enhancement would have an extension to break which would work thus:
break; // our trusty friend, breaks out of current looping construct.
break 2; // breaks out of the current and it's parent looping construct.
break 3; // breaks out of 3 looping constructs.
break all; // totally decimates any looping constructs in force.
I think unless you want to do the "boolean thing" the only solution is actually to throw. Which you obviously shouldn't do..!
Is it possible to refactor the nested for loop into a private method? That way you could simply 'return' out of the method to exit the loop.
Throw a custom exception which goes out outter loop.
It works for for,foreach or while or any kind of loop and any language that uses try catch exception block
try
{
foreach (object o in list)
{
foreach (object another in otherList)
{
// ... some stuff here
if (condition)
{
throw new CustomExcpetion();
}
}
}
}
catch (CustomException)
{
// log
}
As i see you accepted the answer in which the person refers you goto statement, where in modern programming and in expert opinion goto is a killer, we called it a killer in programming which have some certain reasons, which i will not discuss it over here at this point, but the solution of your question is very simple, you can use a Boolean flag in this kind of scenario like i will demonstrate it in my example:
for (; j < 10; j++)
{
//solution
bool breakme = false;
for (int k = 1; k < 10; k++)
{
//place the condition where you want to stop it
if ()
{
breakme = true;
break;
}
}
if(breakme)
break;
}
simple and plain. :)
Since I first saw break in C a couple of decades back, this problem has vexed me. I was hoping some language enhancement would have an extension to break which would work thus:
break; // our trusty friend, breaks out of current looping construct.
break 2; // breaks out of the current and it's parent looping construct.
break 3; // breaks out of 3 looping constructs.
break all; // totally decimates any looping constructs in force.
That's how I did it. Still a workaround.
foreach (var substring in substrings) {_x000D_
//To be used to break from 1st loop._x000D_
int breaker=1;_x000D_
foreach (char c in substring) {_x000D_
if (char.IsLetter(c)) {_x000D_
Console.WriteLine(line.IndexOf(c));_x000D_
\\setting condition to break from 1st loop._x000D_
breaker=9;_x000D_
break;_x000D_
}_x000D_
}_x000D_
if (breaker==9) {_x000D_
break;_x000D_
}_x000D_
}Don't quote me on this, but you could use goto as suggested in the MSDN. There are other solutions, as including a flag that is checked in each iteration of both loops. Finally you could use an exception as a really heavyweight solution to your problem.
GOTO:
for ( int i = 0; i < 10; ++i ) {
for ( int j = 0; j < 10; ++j ) {
// code
if ( break_condition ) goto End;
// more code
}
}
End: ;
Condition:
bool exit = false;
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
exit = true;
break; // or continue
}
// more code
}
}
Exception:
try {
for ( int i = 0; i < 10 && !exit; ++i ) {
for ( int j = 0; j < 10 && !exit; ++j ) {
// code
if ( break_condition ) {
throw new Exception()
}
// more code
}
}
catch ( Exception e ) {}
C# adaptation of approach often used in C - set value of outer loop's variable outside of loop conditions (i.e. for loop using int variable INT_MAX -1 is often good choice):
for (int i = 0; i < 100; i++)
{
for (int j = 0; j < 100; j++)
{
if (exit_condition)
{
// cause the outer loop to break:
// use i = INT_MAX - 1; otherwise i++ == INT_MIN < 100 and loop will continue
i = int.MaxValue - 1;
Console.WriteLine("Hi");
// break the inner loop
break;
}
}
// if you have code in outer loop it will execute after break from inner loop
}
As note in code says break will not magically jump to next iteration of the outer loop - so if you have code outside of inner loop this approach requires more checks. Consider other solutions in such case.
This approach works with for and while loops but does not work for foreach. In case of foreach you won't have code access to the hidden enumerator so you can't change it (and even if you could IEnumerator doesn't have some "MoveToEnd" method).
Acknowledgments to inlined comments' authors:
i = INT_MAX - 1 suggestion by Meta
for/foreach comment by ygoe.
Proper IntMax by jmbpiano
remark about code after inner loop by blizpasta
Use a suitable guard in the outer loop. Set the guard in the inner loop before you break.
bool exitedInner = false;
for (int i = 0; i < N && !exitedInner; ++i) {
.... some outer loop stuff
for (int j = 0; j < M; ++j) {
if (sometest) {
exitedInner = true;
break;
}
}
if (!exitedInner) {
... more outer loop stuff
}
}
Or better yet, abstract the inner loop into a method and exit the outer loop when it returns false.
for (int i = 0; i < N; ++i) {
.... some outer loop stuff
if (!doInner(i, N, M)) {
break;
}
... more outer loop stuff
}
I think unless you want to do the "boolean thing" the only solution is actually to throw. Which you obviously shouldn't do..!
The cleanest, shortest, and most reusable way is a self invoked anonymous function:
One line shorter than the top answer with anonymous method.
new Action(() =>
{
for (int x = 0; x < 100; x++)
{
for (int y = 0; y < 100; y++)
{
return; // exits self invoked lambda expression
}
}
})();
Console.WriteLine("Hi");
Is it possible to refactor the nested for loop into a private method? That way you could simply 'return' out of the method to exit the loop.
factor into a function/method and use early return, or rearrange your loops into a while-clause. goto/exceptions/whatever are certainly not appropriate here.
def do_until_equal():
foreach a:
foreach b:
if a==b: return
Is it possible to refactor the nested for loop into a private method? That way you could simply 'return' out of the method to exit the loop.
This solution does not apply to C#
For people who found this question via other languages, Javascript, Java, and D allows labeled breaks and continues:
outer: while(fn1())
{
while(fn2())
{
if(fn3()) continue outer;
if(fn4()) break outer;
}
}
Did you even look at the break keyword? O.o
This is just pseudo-code, but you should be able to see what I mean:
<?php
for(...) {
while(...) {
foreach(...) {
break 3;
}
}
}
If you think about break being a function like break(), then it's parameter would be the number of loops to break out of. As we are in the third loop in the code here, we can break out of all three.
Manual: http://php.net/break
I've seen a lot of examples that use "break" but none that use "continue".
It still would require a flag of some sort in the inner loop:
while( some_condition )
{
// outer loop stuff
...
bool get_out = false;
for(...)
{
// inner loop stuff
...
get_out = true;
break;
}
if( get_out )
{
some_condition=false;
continue;
}
// more out loop stuff
...
}
As i see you accepted the answer in which the person refers you goto statement, where in modern programming and in expert opinion goto is a killer, we called it a killer in programming which have some certain reasons, which i will not discuss it over here at this point, but the solution of your question is very simple, you can use a Boolean flag in this kind of scenario like i will demonstrate it in my example:
for (; j < 10; j++)
{
//solution
bool breakme = false;
for (int k = 1; k < 10; k++)
{
//place the condition where you want to stop it
if ()
{
breakme = true;
break;
}
}
if(breakme)
break;
}
simple and plain. :)
The easiest way to end a double loop would be directly ending the first loop
string TestStr = "The frog jumped over the hill";
char[] KillChar = {'w', 'l'};
for(int i = 0; i < TestStr.Length; i++)
{
for(int E = 0; E < KillChar.Length; E++)
{
if(KillChar[E] == TestStr[i])
{
i = TestStr.Length; //Ends First Loop
break; //Ends Second Loop
}
}
}
I think unless you want to do the "boolean thing" the only solution is actually to throw. Which you obviously shouldn't do..!
Loops can be broken using custom conditions in the loop, allowing as to have clean code.
static void Main(string[] args)
{
bool isBreak = false;
for (int i = 0; ConditionLoop(isBreak, i, 500); i++)
{
Console.WriteLine($"External loop iteration {i}");
for (int j = 0; ConditionLoop(isBreak, j, 500); j++)
{
Console.WriteLine($"Inner loop iteration {j}");
// This code is only to produce the break.
if (j > 3)
{
isBreak = true;
}
}
Console.WriteLine("The code after the inner loop will be executed when breaks");
}
Console.ReadKey();
}
private static bool ConditionLoop(bool isBreak, int i, int maxIterations) => i < maxIterations && !isBreak;
With this code we ontain the following output:
The easiest way to end a double loop would be directly ending the first loop
string TestStr = "The frog jumped over the hill";
char[] KillChar = {'w', 'l'};
for(int i = 0; i < TestStr.Length; i++)
{
for(int E = 0; E < KillChar.Length; E++)
{
if(KillChar[E] == TestStr[i])
{
i = TestStr.Length; //Ends First Loop
break; //Ends Second Loop
}
}
}
factor into a function/method and use early return, or rearrange your loops into a while-clause. goto/exceptions/whatever are certainly not appropriate here.
def do_until_equal():
foreach a:
foreach b:
if a==b: return
You asked for a combination of quick, nice, no use of a boolean, no use of goto, and C#. You've ruled out all possible ways of doing what you want.
The most quick and least ugly way is to use a goto.
factor into a function/method and use early return, or rearrange your loops into a while-clause. goto/exceptions/whatever are certainly not appropriate here.
def do_until_equal():
foreach a:
foreach b:
if a==b: return
It seems to me like people dislike a goto statement a lot, so I felt the need to straighten this out a bit.
I believe the 'emotions' people have about goto eventually boil down to understanding of code and (misconceptions) about possible performance implications. Before answering the question, I will therefore first go into some of the details on how it's compiled.
As we all know, C# is compiled to IL, which is then compiled to assembler using an SSA compiler. I'll give a bit of insights into how this all works, and then try to answer the question itself.
From C# to IL
First we need a piece of C# code. Let's start simple:
foreach (var item in array)
{
// ...
break;
// ...
}
I'll do this step by step to give you a good idea of what happens under the hood.
First translation: from foreach to the equivalent for loop (Note: I'm using an array here, because I don't want to get into details of IDisposable -- in which case I'd also have to use an IEnumerable):
for (int i=0; i<array.Length; ++i)
{
var item = array[i];
// ...
break;
// ...
}
Second translation: the for and break is translated into an easier equivalent:
int i=0;
while (i < array.Length)
{
var item = array[i];
// ...
break;
// ...
++i;
}
And third translation (this is the equivalent of the IL code): we change break and while into a branch:
int i=0; // for initialization
startLoop:
if (i >= array.Length) // for condition
{
goto exitLoop;
}
var item = array[i];
// ...
goto exitLoop; // break
// ...
++i; // for post-expression
goto startLoop;
While the compiler does these things in a single step, it gives you insight into the process. The IL code that evolves from the C# program is the literal translation of the last C# code. You can see for yourself here: https://dotnetfiddle.net/QaiLRz (click 'view IL')
Now, one thing you have observed here is that during the process, the code becomes more complex. The easiest way to observe this is by the fact that we needed more and more code to ackomplish the same thing. You might also argue that foreach, for, while and break are actually short-hands for goto, which is partly true.
From IL to Assembler
The .NET JIT compiler is an SSA compiler. I won't go into all the details of SSA form here and how to create an optimizing compiler, it's just too much, but can give a basic understanding about what will happen. For a deeper understanding, it's best to start reading up on optimizing compilers (I do like this book for a brief introduction: http://ssabook.gforge.inria.fr/latest/book.pdf ) and LLVM (llvm.org).
Every optimizing compiler relies on the fact that code is easy and follows predictable patterns. In the case of FOR loops, we use graph theory to analyze branches, and then optimize things like cycli in our branches (e.g. branches backwards).
However, we now have forward branches to implement our loops. As you might have guessed, this is actually one of the first steps the JIT is going to fix, like this:
int i=0; // for initialization
if (i >= array.Length) // for condition
{
goto endOfLoop;
}
startLoop:
var item = array[i];
// ...
goto endOfLoop; // break
// ...
++i; // for post-expression
if (i >= array.Length) // for condition
{
goto startLoop;
}
endOfLoop:
// ...
As you can see, we now have a backward branch, which is our little loop. The only thing that's still nasty here is the branch that we ended up with due to our break statement. In some cases, we can move this in the same way, but in others it's there to stay.
So why does the compiler do this? Well, if we can unroll the loop, we might be able to vectorize it. We might even be able to proof that there's just constants being added, which means our whole loop could vanish into thin air. To summarize: by making the patterns predictable (by making the branches predictable), we can proof that certain conditions hold in our loop, which means we can do magic during the JIT optimization.
However, branches tend to break those nice predictable patterns, which is something optimizers therefore kind-a dislike. Break, continue, goto - they all intend to break these predictable patterns- and are therefore not really 'nice'.
You should also realize at this point that a simple foreach is more predictable then a bunch of goto statements that go all over the place. In terms of (1) readability and (2) from an optimizer perspective, it's both the better solution.
Another thing worth mentioning is that it's very relevant for optimizing compilers to assign registers to variables (a process called register allocation). As you might know, there's only a finite number of registers in your CPU and they are by far the fastest pieces of memory in your hardware. Variables used in code that's in the inner-most loop, are more likely to get a register assigned, while variables outside of your loop are less important (because this code is probably hit less).
Help, too much complexity... what should I do?
The bottom line is that you should always use the language constructs you have at your disposal, which will usually (implictly) build predictable patterns for your compiler. Try to avoid strange branches if possible (specifically: break, continue, goto or a return in the middle of nothing).
The good news here is that these predictable patterns are both easy to read (for humans) and easy to spot (for compilers).
One of those patterns is called SESE, which stands for Single Entry Single Exit.
And now we get to the real question.
Imagine that you have something like this:
// a is a variable.
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a)
{
// break everything
}
}
}
The easiest way to make this a predictable pattern is to simply eliminate the if completely:
int i, j;
for (i=0; i<100 && i*j <= a; ++i)
{
for (j=0; j<100 && i*j <= a; ++j)
{
// ...
}
}
In other cases you can also split the method into 2 methods:
// Outer loop in method 1:
for (i=0; i<100 && processInner(i); ++i)
{
}
private bool processInner(int i)
{
int j;
for (j=0; j<100 && i*j <= a; ++j)
{
// ...
}
return i*j<=a;
}
Temporary variables? Good, bad or ugly?
You might even decide to return a boolean from within the loop (but I personally prefer the SESE form because that's how the compiler will see it and I think it's cleaner to read).
Some people think it's cleaner to use a temporary variable, and propose a solution like this:
bool more = true;
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a) { more = false; break; } // yuck.
// ...
}
if (!more) { break; } // yuck.
// ...
}
// ...
I personally am opposed to this approach. Look again on how the code is compiled. Now think about what this will do with these nice, predictable patterns. Get the picture?
Right, let me spell it out. What will happen is that:
more variable that only happens to be used in control flow. more will be eliminated from the program, and only branches remain. These branches will be optimized, so you will get only a single branch out of the inner loop.more is definitely used in the inner-most loop, so if the compiler won't optimize it away, it has a high chance to be allocated to a register (which eats up valuable register memory).So, to summarize: the optimizer in your compiler will go into a hell of a lot of trouble to figure out that more is only used for the control flow, and in the best case scenario will translate it to a single branch outside of the outer for loop.
In other words, the best case scenario is that it will end up with the equivalent of this:
for (int i=0; i<100; ++i)
{
for (int j=0; j<100; ++j)
{
// ...
if (i*j > a) { goto exitLoop; } // perhaps add a comment
// ...
}
// ...
}
exitLoop:
// ...
My personal opinion on this is quite simple: if this is what we intended all along, let's make the world easier for both the compiler and readability, and write that right away.
tl;dr:
Bottom line:
goto or bool more, prefer the former.I remember from my student days that it was said it's mathematically provable that you can do anything in code without a goto (i.e. there is no situation where goto is the only answer). So, I never use goto's (just my personal preference, not suggesting that i'm right or wrong)
Anyways, to break out of nested loops I do something like this:
var isDone = false;
for (var x in collectionX) {
for (var y in collectionY) {
for (var z in collectionZ) {
if (conditionMet) {
// some code
isDone = true;
}
if (isDone)
break;
}
if (isDone)
break;
}
if (isDone)
break;
}
... i hope that helps for those who like me are anti-goto "fanboys" :)
Source: Stackoverflow.com