Martin Fowler, ThoughtWorks

In the early days of corporate computing, application silos were commonplace. As businesses became more sophisticated in their use of IT for competitive advantage, these silos became a bottleneck that prevented seamless straight-through processing and inhibited change.

To break down the silos, at first rudimentary and then increasingly sophisticated integration middleware came to market with the promise of freeing data from the tyranny of the silo. Over the years enterprise middleware has come to dominate corporate computing, and knowledge of such systems is part of every enterprise architect's toolkit.

Yet with the emergence of the Web as a scalable platform for connectivity, and the increasing reluctance for businesses to engage in such large-scale integration projects the future for both "enterprise" and "middleware" is anything but certain. Indeed with the Web and agile methods becoming accepted at scale and quality of service far in excess of most enterprises, "enterprise-y" has become a by-word for backward.

In this keynote, Martin and Jim will explore the history of integration middleware and take a trip into the near future where the Internet has collided with the enterprise providing scale and robustness as a ready commodity, and where agility is prized above all.

Domain Specific Languages (DSLs) are an old technique in software development that's getting a recent resurgence in interest. Most developers run into them regularly - as XML configuration files, regular expressions, query languages or build scripts. However they haven't been given the attention they deserve and there is very little information out there to help developers build them effectively. We find that few people have done much to build their own DSLs and even fewer have a broad appreciation of the various techniques involved.

This tutorial is a step towards closing this gap. We'll begin by introducing the three main categories of DSLs: External, Internal, and Language Workbenches. We'll talk about the advantages of DSLs and the problems in using them, so that you'll appreciate what the different styles look like and when you might want to build them. In the second part we'll go into more details on techniques of working with each of the three styles, to get you started on your own work.

We are currently working to develop a coherent pedagogic framework (if you'll forgive a pretentious name) for DSLs, this tutorial is an opportunity to catch up with our work. However it does come with a caveat: we are still very much in the middle of the process of capturing and organizing this knowledge. As a result we won't be describing a finished body of knowledge, but rather one that is still evolving.