Microsoft XML Architect Don Box signed on with Microsoft in January, 2002, after a long career of consulting and writing about Microsoft from the outside. Box co-founded the software think tank and training company DevelopMentor. He is the author of several books on Microsoft's Component Object Model (COM), one of the precursors to Simple Object Access Protocol (SOAP), XML and .Net. He also is one of a handful of authors of the original SOAP specification.

The free-spirited Box dabbles in fashion, with a line of underwear called "Don Boxers." He's also chief songwriter, singer and guitarist for "Band on the Runtime," a group of rock-playing geeks. At the recent Professional Developers Conference (PDC), Box didn't engage in any of his trademark speaking stunts, such as wearing his boxers on stage, or having a back drop of soap bubbles (unless you count calling Group VP Jim Allchin "code monkey," during one of the keynote addresses as something other than a career-endangering move).

Check Out Don Box's 'Spoutlet' Weblog

And a Recap of the Band on the Runtime's PDC Gig for the Press

Box has been coding like a madman ever since the PDC. So we never got time with him live. But here's our e-mail Q&A with one of the main movers and shakers behind the "Indigo" communications subsystem due to be part of Longhorn.

Microsoft Watch: What is your day-to-day job these days?

Box: I'm one of several architects on the Indigo team. When I joined the Indigo team in early 2002, my focus was on the protocol stack - primarily working on the architecture and core WS-* specs like WS-Addressing and WS-Policy. Around the beginning of the year, my attention turned towards the product, specifically working on our XML infrastructure and helping define our service-oriented programming model. While I spend several hours a week coveting the stuff the transactions team is doing, my contributions there have been fairly minor to date.

What Is Indigo?

Microsoft Watch: In a nutshell, how has your transition to Microsoft gone? High points? Low points? Are you glad you joined "the Empire"?

Box: Yes, I'm glad I made the move. There's no shortage of interesting problems to work on and more importantly, there's an abundance of smart people to work with and learn from. The most frustrating aspect of the move is the long ship cycles many of our products are on. This frustration is real if not completely rational — given the size of some of the bets we're making (Avalon, WinFS, Indigo) it's hard to imagine shipping something the scale of Longhorn much faster.

Microsoft Watch: How did the idea for Indigo come about? Were you part of it? How/when did communications (P2P; video chat, etc.) come to be integrated/added to the Indigo vision?

Box: Indigo was conceived in 2000 to do for services what the CLR (Common Language Runtime) did for objects. Just as most services are constructed using object technology, Indigo is a heavy user of the CLR. When I found out about Indigo in 2002, I was blown away by the breadth of the commitment so I decided it was time to move to Redmond. Ultimately, Indigo is about getting messages into and out of a CLR AppDomain as quickly as possible, so to that end, I expect more connected applications to build on Indigo over time.

Microsoft Watch: What's the future of SOAP, in your opinion (and as one of the original authors of the SOAP spec)? Is it evolving into Service Oriented Architecture Protocol than Simple Object Access Protocol? (and are these two things different)?

Box: In 1998, I don't think any of us anticipated the level of adoption (and hype) we've seen around SOAP-based technologies. It's now more than five years later and I can say confidently that SOAP itself is done. The WS-I Basic Profile cleaned up some of our ambiguities in SOAP/1.1 (the version of SOAP that is most widely deployed today) and the W3C finished the SOAP/1.2 specification, which gives the world a stable base to build upon going forward. The evolution of SOAP is no longer about SOAP itself but rather what people do with SOAP. Yes, the earliest application of SOAP was RPC (remote-procedure call). Going forward, RPC plays a role but isn't necessarily the primary way people will build SOAP-based systems. As for SOAP's acronym status, it was officially revoked in the SOAP/1.2 spec by my good friend Martin Gudgin.

Microsoft Watch: On to the latest buzzword: Service Oriented Architectures, a k a SOAs. Could you give a lay person's definition? Are SOAs really anything more than a message-based RPC mechanism?

Box: Given that seemingly every piece of integration software/middleware claims to be based on SOA, it would be presumptuous for me to define it for the entire industry. What I can say is what we're doing with services in Indigo. Indigo makes service-oriented programming approachable (and real) to developers who use the CLR. We're providing runtime, framework, and tool-level support for making the boundaries between services explicit without introducing complexity into the programming model. When compared to object-orientation, service-oriented programming is most conspicuous in what it leaves out, not what it adds.

Microsoft Watch: Do you foresee the service-oriented paradigm replacing the object-oriented paradigm? Explain.

Box: They're for different purposes. For software that is tested, deployed, and versioned as an atomic unit, object technology will continue to be the dominant way people write programs. However, we're seeing an industry-wide move away from using objects as the metaphor we use when integrating autonomous programs that are tested, deployed and versioned independently. For that, services appear to be the right answer.

Microsoft Watch: Are you going to be doing anything at Microsoft for/with the XML Application Markup Language (XAML) team?

Read More on XAML (and Sparkle)

Box: I like XAML a lot. Now that PDC is behind us, I'm spending more time with XAML and the people who own it.

Microsoft Watch: Whatever happened to X#? (Editor's Note: X# is the code name for an XML programming language that Microsoft proposed to develop. Box mentioned it a year ago at the XML 2002 conference.)

For the Original X# Report, Read This

Box: The ideas put forth by X# have influenced a lot of people inside of Microsoft. Beyond that, I'm not plugged into the product plans around X#.

Microsoft Watch: What do you want with Miguel de Icaza? Why are you interested in seeing him join Microsoft?

Box: Miguel has an amazing amount of raw energy. I for one would like to see that energy applied to the initial design and architecture phase of a technology rather than on paying homage to work that's already been done.

Microsoft Watch: Do you foresee your working partnership with IBM around Web services continuing? Will there be a WS roadmap part 2? Or have you two established enough momentum with roadmap 1?

Check Out the Roadmap 1 for a Refresher

Box: We now have a basis for secure, reliable, transactions in place that we need to drive to completion. This means keeping the implement/interop/refine cycle moving as smoothly and quickly as possible. How much more common functionality is needed is hard to say, as a lot of what's left is more of an application of what we've already done.

Microsoft Watch: Do you believe that Web services players like MS need to support all of a WS protocol in its entirety? Specifically: Is MS supporting all of WS-I Basic Profile? If not, can there still be a guarantee of Web services interoperability?

Box: Yes, we're supporting WS-I Basic Profile 1.0. Our shipping product already allows developers to create BP 1.0-compliant services and we publish guidelines to help developers stay within the BP 1.0 boundaries.

(This interview originally appeared in the November 20, 2003, issue of the Microsoft Watch newsletter.)