A lot of the discussion about .NET 4.0 is revolving around the introduction of co-variance and contra-variance for generic types.

It’s important to remember, though, that these concepts aren’t entirely new - there has been some support for variance built into .NET for some time.

For example, consider this snippet of code:

    public Form1()
    {
        InitializeComponent();
        textBox1.KeyPress += Handler;
        button1.Click += Handler;
        textBox1.KeyPress += new KeyPressEventHandler(Handler);
        button1.Click += new EventHandler(Handler);
    }

    private void Handler(object sender, EventArgs e)
    {
        label1.Visible = true;
    }

You can see here the Handler() method being registered for both the TextBox.KeyPress event and the Button.Click event, even though these events have different parameters (Button.Click expects parameters object and EventArgs while TextBox.KeyPress expects object and KeyPressEventArgs).

What’s going on?

Without specific compiler support, this wouldn’t work. But, variance comes into play. The compiler recognises that the second parameter of Handler() is of type EventArgs - which is sufficiently general to accept any possible value that might be supplied by the event. Based on this observation, the compiler hooks up the event without complaint.

Comments

blog comments powered by Disqus
Next Post
TechEd New Zealand 2009  19 Sep 2009
Prior Post
Coming up to speed  10 Sep 2009
Related Posts
Short-circuiting validation  11 Aug 2018
Aggregation of validation  04 Aug 2018
Recovery of validation types  28 Jul 2018
Basic validation  21 Jul 2018
Why we need better validation  07 Jul 2018
A question of struct performance  30 Jun 2018
Equality has Symmetry  17 Mar 2018
Type Miscellany  05 Mar 2017
Testing Immutable Types  25 Feb 2017
Factory Methods  18 Feb 2017
More csharp posts »
Archives
September 2009
2009