I’ve just returned from the PDC in LA and have seen a number of news wire articles and analyst comments being somewhat negative about Microsoft delivering too little too late. These remarks have generally been directed at Longhorn and the fact that it is going to arrive later with fewer features than originally planned.
I’ve been to a few PDCs so I know it’s easy to get caught up in all the razzmatazz and glitz of the event, but even so, I think that Microsoft have moved along away forward across an extremely broad set of technologies. We (Conchango) have been exposed to the technology and features within .Net 2.0, SQL 2005 and Visual Studio & Team System 2005 through our project work, and to some extent have become acclimatised to this step change. Despite this preparation I was still blown away by the power of WinFX which comprises a set of major components and their managed .Net APIs. It was the announcement of Windows Workflow Foundation that really got my attention, and not just because Don Box gave typically charismatic performances when presenting! Not even a codename this time as Microsoft have kept it pretty quiet with just a few unsubstantiated rumours floating around the blogosphere.
I’ve always struggled with the concept of bolting an external workflow engine on top of an application(s). It isn’t generally very seamless and it’s hard to cleanly separate out the workflow from the application logic. WWF (hopefully not at all like wrestling!), should allow us to build application workflow and orchestrate web services from within the .Net Framework and IDE, but still allow the abstraction of workflow, exposing it so that its less opaque and more easily modified post compilation. I’m also pleased to see that Microsoft have chosen to support, not only the normal predefined fixed flow model, but also a model they call the ‘State Machine’ workflow which looks more event driven, supporting an indeterminate execution order. This event driven model will enable the composition of services and business processes in to a highly flexible animal, which can much more accurately match the variable and unpredictable nature of the reality of business today. WWF should also further encourage people away from monolithic approaches to building applications, towards a more granular service based model, that can be flexibly wrapped by WWF. Discrete business functions can be built as services and then consumed and co-opted by any number of business flows built in WWF.
Another potential and extremely interesting possibility is to use WWF as a page flow controller in web applications. This would be fantastic if it turns out to be fast and scalable enough and would be a lot neater than coding up an MVC pattern every time you build an application. Page to page flow could be extremely flexible and you would be able to modify it without recompiling the code. I think we might have to put some prototypes together!