In our premier issue, back in October 2002, we ran a full-length interview with Anders Hejlsberg, the Distinguished Engineer at Microsoft responsible for the creation of the C# programming language. Then, in March 2005, we presented a follow-up interview conducted by .NETDJ's editor-in-chief, Derek Ferguson, at Microsoft's Tech Ed 2004 conference in San Diego, California. Here it is again in full.

.NETDJ: Between now and the last time we spoke, Borland has entered the .NET space. As an ex-Borland employee who is now one of the most revered .NET icons, what are your thoughts on this?

AH: I'm very excited that they are doing that! It is the right thing for the Delphi community and .NET is the logical next place for all Windows development tools to go. I think it is a win-win situation.

.NETDJ: I didn't realize the last time we spoke that you started off at Microsoft building the Windows Foundation Classes for J++. I never had a chance to look at WFC during its day. If I looked at it today, would I see bits I'd recognize from the .NET Framework?

AH: Some of the ideas from WFC were carried forward into Windows Forms, so you would see stuff there.

.NETDJ: Where, if anywhere, do you see a need for "a language specializing in... <blank>" today? For example, I think there is a real need for a new entry-level programming language. What do you think?

AH: I think there is a healthy cross-pollination that occurs between all the programming languages. While you might see C# pioneer in one direction and VB pioneer another, ultimately there is a lot of crossing over of ideas over time. Certainly, anything we do on the C# team we let the VB folks participate in and vice versa.

.NETDJ: I recently spoke with Soma Somasegar, Microsoft's Corporate VP of the Developer Division, and he mentioned that you were one of the people looking at how XML usage can be done better from a programmatic standpoint. Can you tell us a little about this work?

AH: Well, I wasn't involved in X#, but two of the people who were on the X# team are now members of the C# team. Some of X#'s ideas are being carried forward in C#. These are not going to look the same way as they did in X#. However, I think that the core concepts will carry forward. I can't make any commitments, though.

Actually, the main thrust of X# was trying to deal with relational data. It also tried to deal with XML, but it did not deal with loosely typed XML, which is what most of the XML in the world is.

My feeling is that if XML is strongly typed, then it can very conveniently be represented as classes. So what we really need is syntax for creating "object literals" - objects with nested objects - all in a single expression. We are kind of calling these "object initializers."

I would not expect to see XML literals embedded in the C# programming language, however. There are two ways to think about integrating concepts. On one hand, you can just embed one language in another. Honestly, however, this approach just calls attention immediately to all of the differences between the two languages. Take Embedded SQL, for example. You still have to learn two languages, but you also have to learn how those two languages interact.

On the other hand, if you could just take the conceptual things that are in two languages and merge them into one, you could get a much more productive environment. This is the approach to data I want to take in future versions of C#.

.NETDJ: The last time we spoke (at OOPSLA 2002), you had just announced that anonymous methods would be added to C#, but you couldn't give a timeline. Now we know that that will occur in Visual Studio 2005. What has happened between now and then to take this idea from a vision to reality?

AH: The stuff we talked about at OOPSLA was the first glimpse we gave of the new C# language features that will be in Visual Studio 2005. There are many that are well-known: generics, anonymous methods, iterators, and partial classes. There are also many more - nullable types, which we just talked about today - and a whole bunch of smaller things that are not as well-known.

Nullable types basically represent the ability to have value types that can be set to null, which is something that people very much ask for. I recently asked a group of programmers, "How many of you access data in your applications?" Virtually everyone raised his hand.

People do a lot of database access, and yet there are some pretty notable impediments to interactions between databases and programming languages. One of these mismatches is the absence of nullable types from programming languages. SQL databases have always had nullable types, but programming languages have never had nullable types. One of the things that you are sort of seeing here is the beginning of us trying to think really deeply about this problem. Nullable types are a specific problem that we are solving as a part of .NET 2.0. In .NET 1.0, value types can't be null, but in 2.0 they can.

(continued on page 2)