As I get ready to celebrate the holidays this year, I spent a little time reflecting on what I would like from the various Web services vendors. While a case of scotch was definitely tempting, what I really want is a better toolset.

To date, most of the programming tools available for building Web services have focused on the ultimate deployment platform. In other words, they've looked at the end implementation language first and tuned their work towards that. So, for example, Visual Studio aims at C# or VB, WebSphere Application Developer aims at WebSphere, and BEA WebLogic Workshop targets WebLogic. Borland, who also makes generally excellent tools, has built support into JBuilder for Web services, but again, targeting Java.

Now don't get me wrong. We need tools that will allow us to focus our energy on implementing Web services in particular languages. After all, a service is worthless unless you can actually invoke it. So these tools are all very important parts of the overall Web services development strategy.

At the same time, they've all focused on the only platform/language/idiom-specific aspect of the entire Web services spectrum - the implementation. Every other aspect of Web services is vendor, language, and platform neutral. And there are no tools for that.

I want a tool that makes it simple for me to first define my Web service, then pass it over to whatever development environment I choose for implementation. And when you look at it from that perspective, you see what I mean.

I want something that makes it easy to define the service so that I can visually create a WSDL file. I want something that understands more than just SOAP, WSDL, and UDDI - I want the tool aware of and capable of managing and integrating WS-Orchestration, WS-Security, BPEL, and all the other standards. I want it to be able to conform a previously defined service to these standards (so that I can make something simple more complex, like adding transactionality) without having to burn incense and read three product manuals.

Ideally, we can use this for more than just a better "developer's" tool as well. Much of the software development process these days comes long before the actual coding - things like requirements gathering and analysis, design, and modeling all have their place in the modern arena of software development. Elements of each of those domains should also be part of this toolset - it should be possible to do an enterprise API with this tool, and then get more granular and refined as we move ideas closer to reality.

Of course, the ultimate goal of this tool should be to produce a Web service stub for many different target platforms. As I mentioned, a service is worthless if no one can use it because it can't be implemented. Likewise, all of this language-independent work is meaningless if it can't be directly applied to a particular implementation environment. That was always a problem with the modeling tools in the past - round-trip engineering always came last, and always had bugs. Fortunately, in this case most of the information we provide, right up to the actual code, resides in XML, which can be modeled and transformed quite easily.

So there's my hope for a happy New Year - a tool that makes it easier to do the front-end work - the design and definition of Web services, before having to actually code them. Of course, I'll take that case of scotch if it comes too! Happy holidays.