You'll be glad to know! [modified]

[PIEBALDconsult]

Until I did some refactoring today I had a method with three return statements! :omg: (They were to avoid a goto to abort a nested loop.)

[Luc Pattyn]

Well, the complexity of avar1 = someCalculation(); isn't relevant, you can do it upfront anyway. The complexity of avar2 = someOtherCalculation(); is what matters. If it can't be put in the middle of the if clause, I'd use a boolean flag, like so:

for (...) {
bool updateWanted=false;
if (calculation()==someValue) updateWanted=true;
else if (someOtherCalculation()==someValue2) updateWanted=true;
if (updateWanted) localVar.UpdateNicely();
}

and now you can easily expand line 4 (and 3) as required. :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, and improve readability.

modified on Thursday, September 9, 2010 8:52 PM

[AspDotNetDev]

PIEBALDconsult wrote:

avoid a goto to abort a nested loop

I usually set some bool, break, then conditionally break depending on the value in the bool. Slows things down a bit, so a goto might be more acceptable in this case if speed is of significant importance and the inner loop does not have many iterations.

[Forum Guidelines]

[Luc Pattyn]

it's like writing poetry, just keep scrapping the redundant bits until you start to like the result. Except you have to adhere to the original intentions... :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, and improve readability.

[Super Lloyd]

see the code below! it can't be done upfront, as the first if will avoid as divide by zero in the second if! just for your benefit I post the code below again! Anyway, yes it can be done, but I found my way more readable... I guess it's too subjective for debate.. but I invite you to show me the new version:

for (int i = max; i >= min; i--)
{
var v1 = new Vector2D(points[i > 0 ? i - 1 : points.Count - 1], points[i]);
var v2 = new Vector2D(points[i], points[i < points.Count - 1 ? i + 1 : 0]);
if (v1.SquareNorm <= MINL || v2.SquareNorm <= MINL)
goto RemovePoint;
v1 = v1.Normalize(); // divide by zero if square norm is 0 (test above)
v2 = v2.Normalize();
var z = v1 ^ v2;
if (Math.Abs(z) <= minsin && v1 * v2 < 0)
goto RemovePoint;
continue;
RemovePoint: ;
points.RemoveAt(i);
if (points.Count < 3)
return null;
}

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Super Lloyd]

Thanks! Glad someone agree! I think I was looking for a beating... but I have a hard time handling it! :laugh:

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Peter_in_2780]

Two comments. 1. Loops are only nested too deep when you need to indent each level by less than [insert favourite small number here, mine's 4] spaces to fit it on the screen. ;P The construct

if (!test)
goto next;
do_stuff;
next:

is actually what a compiler* will emit if you feed it

if (test)
do_stuff;

Ironic, huh? But maybe that's why we deal in the input to compilers and leave their output unseen. * for a typical target instruction set

Software rusts. Simon Stephenson, ca 1994.

[Super Lloyd]

Ho my god! That Is Bad! :laugh:

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[AspDotNetDev]

If you really don't want to put some of that logic into functions, this is one way to go:

for(...)
{
bool doRemove = false;
var v1 = a();
var v2 = b();
if(v1.a() || v2.a())
{
doRemove = true;
}
else
{
v1 = c();
v2 = d();
var z = e();
if(e(z, v1, v2))
{
doRemove = true;
}
}
if(doRemove)
{
RemovePoint();
}
}

I put in some placeholders where I didn't feel like typing.

[Forum Guidelines]

[Super Lloyd]

mmh... so you mean your way is better because it will compile to my way?!? ;P anyway, at this stage it's a matter of taste, your solution is elegant too! but too many bracket confuses me and the compiler in my brain! (as opposed to the one on the hard drive ;) )

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Luc Pattyn]

I already gave my preferred structure here[^], and I think we've had enough code in the Lounge for this week... :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles] Nil Volentibus Arduum

Please use <PRE> tags for code snippets, they preserve indentation, and improve readability.

[Super Lloyd]

ok, I guess this one will work and is readable! :thumbsup: but I still prefer my way! :laugh:

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Super Lloyd]

mm.. I guess it's a nice simplification of other solution and it works AND is easily readable too!! :thumbsup: anyway, I still prefer my code! :laugh:

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Peter_in_2780]

Super Lloyd wrote:

your way is better because it will compile to my way

What I am saying is that it's good for us that compilers can do the messy stuff, leaving us to express our algorithms in ways we can wrap our heads around without spewing. A couple of others have made similar comments in this thread. Enough with the philosophy! I'm outta here! ;P

Software rusts. Simon Stephenson, ca 1994.

[Cesar de Souza]

I've found GOTO support specially useful when porting code from FORTRAN - or even from C. I guess this is one of the real excuses for having a GOTO instruction in C#: dealing with legacy code from languages were GOTO are/were common place. It just becomes much easier to port entire algorithms without having to change its structure in the process. Once an initial port is complete, it is much simpler to remove those offending instructions one by one (specially after you have set lots of regression tests). However, in some cases, removing all gotos just isn't worth the risk.

Interested in Machine Learning in .NET? Check the Accord.NET Framework. See also Handwriting Recognition Revisited: Kernel Support Vector Machines

modified on Saturday, September 11, 2010 9:19 PM

[Marc Clifton]

Super Lloyd wrote:

can give me tips

well, most languages have the following capabilities: "else" methods can call methods "or" operations A combination of one or more of those language features would make your code a hell of a lot more readable and maintainable. ;) Marc

[Saurabh Garg]

This is easy just put doStuff in another function. -Saurabh

[Super Lloyd]

Maybe you an answer this other question. Somewhere I posted my real code and somewhere some posted the nested if version. Personally I found my version way simpler and maintainable. It has less line, less bracket, less variable a lower cyclomatic complexity. Now where does this "absolutely no goto" religion comes from? I just can't bring myself to prefer the nested if extra variable more lines of code version.... :omg: Anyway, I can see this is going no where. I did learn something, it's pointless to post against people belief, nothing came out of it! I haven't change my opinion, nor did they! and I'm upset!

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.

[Steve Echols]

I'm not as much of a hater of goto's as most people, but without goto's I'd write it as:

for (int i = max; i >= min; i--)
{
bool removePoint = false;
var v1 = new Vector2D(points[i > 0 ? i - 1 : points.Count - 1], points[i]);
var v2 = new Vector2D(points[i], points[i < points.Count - 1 ? i + 1 : 0]);
if (v1.SquareNorm <= MINL || v2.SquareNorm <= MINL)
{
removePoint = true;
}
else
{
v1 = v1.Normalize(); // divide by zero if square norm is 0 (test above)
v2 = v2.Normalize();
var z = v1 ^ v2;
if (Math.Abs(z) <= minsin && v1 * v2 < 0)
removePoint = true;
}

    if (removePoint)
    {
        points.RemoveAt(i);
        if (points.Count < 3)
            return null;
    }

}

Either way, it's pretty clear. Yours doesn't require the extra allocation/initialization/assignment of the flag, so it's slightly more efficient, which could be important if you're in a huge loop.

---

- S 50 cups of coffee and you know it's on! Code, follow, or get out of the way.

[Super Lloyd]

Glad to see that not everyone give in the anti goto dogma, an actually look at the code! :-D And yeah, I have been unconvinced by all the gymnastic to get rid of the goto so far. Seems much easier, shorter, etc... with it! I'm not really sure why I made this post in fact.. but I think I learn something about communication! :)

A train station is where the train stops. A bus station is where the bus stops. On my desk, I have a work station.... _________________________________________________________ My programs never have bugs, they just develop random features.