Yield in VB.NET

Viewed 32103

C# has the keyword called yield. VB.NET lacks this keyword. How have the Visual Basic programmers gotten around the lack of this keyword? Do they implement they own iterator class? Or do they try and code to avoid the need of an iterator?

The yield keyword does force the compiler to do some coding behind the scenes. The implementation of iterators in C# and its consequences (part 1) has a good example of that.

8 Answers

Note: This answer is old now. Iterator blocks have since been added to VB.NET

C# translates the yield keyword into a state machine at compile time. VB.NET does not have the yield keyword, but it does have its own mechanism for safely embedding state within a function that is not easily available in C#.

The C# static keyword is normally translated to Visual Basic using the Shared keyword, but there are two places where things get confusing. One is that a C# static class is really a Module in Visual Basic rather than a Shared class (you'd think they'd let you code it either way in Visual Basic, but noooo). The other is that VB.NET does have its own Static keyword. However, Static has a different meaning in VB.NET.

You use the Static keyword in VB.NET to declare a variable inside a function, and when you do the variable retains its state across function calls. This is different than just declaring a private static class member in C#, because a static function member in VB.NET is guaranteed to also be thread-safe, in that the compiler translates it to use the Monitor class at compile time.

So why write all this here? Well, it should be possible to build a re-usable generic Iterator<T> class (or Iterator(Of T) in VB.NET). In this class you would implement the state machine used by C#, with Yield() and Break() methods that correspond to the C# keywords. Then you could use a static instance (in the VB.NET sense) in a function so that it can ultimately do pretty much the same job as C#'s yield in about the same amount of code (discarding the class implemenation itself, since it would be infinitely re-usable).

I haven't cared enough about Yield to attempt it myself, but it should be doable. That said, it's also far from trivial, as C# team member Eric Lippert calls this "the most complicated transformation in the compiler."

I have also come to believe since I wrote the first draft of this over a year ago that it's not really possible in a meaningful way until Visual Studio 2010 comes out, as it would require sending multiple lambdas to the Iterator class and so to be really practical we need .NET 4's support for multi-line lambdas.

I personally just write my own iterator class that inherits from IEnumerator(Of T). It does take some time to get it right, but I think in the end it's better to write it right then try to avoid it. Another method that I have done is to write a recursive method that returns IEnumerable(Of T) and just returns List(Of T) and uses .AddRange.

Hopefully, this will be a thing of the past with the upcoming version of VB. Since iterators are actually gaining a lot of importance with new paradigms (especially LINQ in combination with lazy evaluation), this has quite a high priority, as far as I know from Paul Vick's blog. Then again, Paul's no longer the head of the VB team and I haven't yet had time to watch the PCD talks.

Still, if you're interested, they're linked in Paul's blog.

Related