Issue with Call Hierarchy

Question

I have a situation where I am calling function/method in cascading style. See following example for illustration and question. I wish I knew some technical word for this situation. It would be easier for people understand what I am talking about.

public static class test
{
    private void button1_Click(object sender, RoutedEventArgs e)
    {
        if (Login("johndoe","password")) 
        {
            if(checkForSomething("johndoe"))
            {
                DoOpenDashboard();

                // Now it opens dashboard, it has several buttons. 
                // Each button does several different things
                // On this example I am just giving you two level of hierarchy
                // but in my actual program, there are 7 levels.
            }
        }
    }

    public static bool Login(string userid, string password)
    {
        //valid user
        return true;
    }

    public static bool checkForSomething(string userid) 
    {
        return true;
    }

How do I avoid process to go back to previous calling method/function if child method runs sucessfully?

For example login method is calling checkForSomething("johndoe"). If checkForSomething("johndoe") is passed then it will open Dashboard window by calling DoOpenDashboard. At this point my process should should not go back to checkforsoemthing, and then login. I hope it makes sense.

You cannot have a method with the same name of the class that contains it. — Miguel Angelo, Oct 27 '11 at 20:19
Your question is very difficult to understand. Can you not just add a static variable or instance member to your class to indicate whether or not your user has already logged in, and you only call `Login()` if the value of that boolean is false? — Kiley Naro, Oct 27 '11 at 20:20
Please, move the question outside the code block... it is impossible to read it like it is now. — Miguel Angelo, Oct 27 '11 at 20:21
Even if what you're asking can (easily) be done, you still violate the single responsibility principle by having a method that behaves differently based on how many times you call it. It will certainly result in a difficult to maintain system. — Roman, Oct 27 '11 at 20:25

sq33G · Answer 1 · 2011-10-28T07:43:09.760

Not really clear on what you're asking here. Your pseudocode shows the Login () method being called in your class's constructor. If this really is how your code is working, then to prevent calling Login again you need to avoid creation of new instances of this class.

However, I think you are really asking about the Arrow anti-pattern:

http://codinghorror.com/blog/2006/01/flattening-arrow-code.html

EDIT

I was trying to avoid copy & paste, but since the original post seems not to have been clear enough, here's a selection from Coding Horror as linked above:

Where appropriate, I flatten that arrow code by doing the following:

Replace conditions with guard clauses. This code..

if (SomeNecessaryCondition) { // function body code }

.. works better as a guard clause:

if (!SomeNecessaryCondition) { throw new RequiredConditionMissingException; } // function body code

(note that there are other techniques also listed, but I think that this first one will suffice for now)

This way, each additional check doesn't lead to another nested if - once any check fails, the method call fails. This can also be done without throwing an exception by having button1_Click call a function that returns a bool (true for success, false for failure) and immediately returning false on a failed condition:

private void button1_Click(object sender, RoutedEventArgs e)
{
    if (AllSystemsGo())
    {
        DoOpenDashboard();
    }
}

private bool AllSystemsGo()
{
    if (!Login("johndoe","password"))
        return false;

    if (checkForSomethingEvil("johndoe"))
        return false;

    if (!checkForSomethingImportant())
        return false;

    return true;
}

It's not a constructor, it's just a method with the same name as the class (note that it has a return type). — Roman, Oct 27 '11 at 20:27
It's not obviously one or the other. Obviously this is not compiling code, and your guess is as good as mine whether the return value is a mistake or the method name is a mistake. — sq33G, Oct 27 '11 at 20:40
I am using contion checking as stated in "Flattening Arrow Code" wherever it is possible. I am extracting business logic from the code vb6 code. Since I am extracting and understanding the business logic, i am getting into horriable situation. — Shai, Oct 27 '11 at 21:26

score 0 · Answer 2 · answered Oct 28 '11 at 08:33

Are you trying to make sure that your methods are checked only once? Perhaps you need some properties to be queried multiple times, but tested only once.

private bool? canLogin;
private bool? somethingOk;

private bool CanLogin
{
    get
    {
        if (canLogin == null)
            canLogin = Login("johndoe","password");
        return canLogin.Value;
    }
}

private bool SomethingOk
{
    get
    {
        if (somethingOk == null)
            somethingOk = checkForSomething("johndoe");
        return somethingOk .Value;
    }
}

private void button1_Click(object sender, RoutedEventArgs e)
{
    if (this.CanLogin && this.SomethingOk && // other checks) 
    {
        DoOpenDashboard();            
    }
}

Issue with Call Hierarchy

2 Answers2