PDC 2008 was a lot of fun - a big show, with lots of announcements, lots of sessions and some thought-provoking content. I thought I'd through out a few observations. Not really conclusions, as those take time and reflection, so just some observations.

Windows Azure, the operating system for the cloud, is intriguing. For a first run at this, the technology seems surprisingly complete and contains a pretty reasonable set of features. I can easily see how web sites, XML services and both data-centric and compute-centric processing could be built for this platform. For that matter, it looks like it would be perhaps a week's work to get my web site ported over to run completely in Azure.

The real question is whether that would even make sense, and that comes down to the value proposition. One big component of value is price. Like anyone else, I pay a certain amount to run my web site. Electricity, bandwidth, support time, hardware costs, software costs, etc. I've never really sorted out an exact cost, but it isn't real high on a per-month basis. And I could host on any number of .NET-friendly hosting services that have been around for years, and some of them are pretty inexpensive. So the question becomes whether Azure will be priced in such a way that it is attractive to me. If so, I'm excited about Azure!! If not, then I really don't care about Azure.

I suspect most attendees went through a similar thought process. If Microsoft prices Azure for "the enterprise" then 90% of the developers in the world simply don't care about Azure. But if Microsoft prices Azure for small to mid-size businesses, and for the very small players (like me) then 90% of the developers in the world should (I think) really be looking at this technology

Windows 7 looks good to me. After the Tuesday keynote I was ready to install it now. As time goes by the urgency has faded a bit - Vista has stabilized nicely over the past 6-8 months and I really like it now. Windows 7 has some nice-sounding new features though. Probably the single biggest one is reduced system resource requirements. If Microsoft can deliver on that part of the promise I'll be totally thrilled. Though I really do want multi-monitor RDP and the ability to manage, mount (and even boot from) vhd files directly from the host OS.

In talking to friends of mine that work at Microsoft, my level of confidence in W7 is quite high. A couple of them have been running it for some time now, and while it is clearly pre-beta, they have found it to be a very satisfying experience. When I get back from all my travels I do think I'll have to buy a spare HD for my laptop and give it a try myself.

The Oslo modeling tools are also interesting, though they are more future-looking. Realistically this idea of model-driven development will require a major shift in how our industry thinks about and approaches software development. Such a massive shift will take many years to occur, regardless of whether the technology is there to enable it. It is admirable that Microsoft is taking such a gamble - building a set of tools and technologies for something that might become acceptable to developers in the murky future. Their gamble will pay off if we collectively decide that the world of 3GL development really is at an end and that we need to move to higher levels of abstraction. Of course we could decide to stick with what has (and hasn't) worked for 30+ years, in which case modeling tools will go the way of CASE.

But even if some of the really forward-looking modeling ideas never become palatable, many of the things Microsoft is doing to support modeling are immediately useful. Enhancements to Windows Workflow are a prime example, as is the M language. I've hard a hard time getting excited about WF, because it has felt like a graphical way to do FORTRAN. But some of the enhancements to WF directly address my primary concerns, and I can see myself getting much more interested in WF in the relatively near future. And the ability of the M language to define other languages (create DSLs), where I can create my own output generator to create whatever I need - now that is really, really cool!

Once I get done with my book and all my fall travel, you can bet I'll be exploring the use of M to create a specialized language to simplify the creation of CSLA .NET business classes :)

There were numerous talks about .NET 4.0 and the future of C# and VB.

Probably the single biggest thing on the language front is that Microsoft has finally decided to sync VB and C# so they have feature parity. Enough of this back-and-forth with different features, the languages will now just move forward together. A few years ago I would have argued against this, because competition breeds innovation. But today I don't think it matters, because the innovation is coming from F#, Ruby, Python and various other languages and initiatives. Both VB and C# have such massive pre-existing code-bases (all the code we've written) that they can't move rapidly or explore radical ideas - while some of these other languages are more free to do just that.

The framework itself has all sorts of changes and improvements. I spent less time looking at this than at Azure and Oslo though, so I honestly just don't have a lot to say on it right now. I look at .NET 4.0 and Visual Studio 2010 as being more tactical - things I'll spend a lot of time on over the next few months anyway - so I didn't see so much need to spend my time on it during PDC.

Finally, there were announcements around Silverlight and WPF. If anyone doubts that XAML is the future of the UI on Windows and (to some degree) the web, now is the time to wake up and smell the coffee. I'm obviously convinced Silverlight is going to rapidly become the default technology for building business apps, with WPF and Ajax as fallback positions, and everything at the PDC simply reinforced this viewpoint.

The new Silverlight and WPF toolkits provide better parity between the two XAML dialects, and show how aggressively Microsoft is working to achieve true parity.

But more important is the Silverlight intersection with Azure and Live Mesh. The fact that I can build smart client apps that totally host in Azure or the Mesh is compelling, and puts Silverlight a notch above WPF in terms of being the desired start-point for app development. Yes, I really like WPF, but even if it can host in Azure it probably won't host in Mesh, and in neither case will it be as clean or seamless.

So while I fully appreciate that WPF is good for that small percentage of business apps that need access to DirectX or rich client-side resources, I still think most business apps will work just fine with access to the monitor/keyboard/mouse/memory/CPU provided by Silverlight.

A couple people asked why I think Silverlight is better than Ajax. To me this is drop-dead simple. I can write a class in C# or VB that runs on the client in Silverlight. I can write real smart client applications that run in the browser. And I can run that exact same code on the server too. So I can give the user a very interactive experience, and then re-run that same code on the server because I don't trust the client.

To do that in Ajax you'd either have to write your code twice (in C# and in Javascript), or you'd have to do tons of server calls to simulate the interactivity provided by Silverlight - and that obviously won't scale nearly the same as the more correct Silverlight solution.

To me it is a no-brainer - Ajax loses when it comes to building interactive business apps like order entry screens, customer maintenance screens, etc.

That's not to say Ajax has no home. The web and browser world is really good at displaying data, and Ajax makes data display more interesting that simple HTML. I strongly suspect that most "Silverlight" apps will make heavy use of HTML/Ajax for data display, but I just can't see why anyone would willingly choose to create data entry forms or other interactive parts of their app outside of Silverlight.

And that wraps up my on-the-flight-home summary of thoughts about PDC.

Next week I'm speaking at the Patterns and Practices Summit in Redmond, and then I'll be at Tech Ed EMEA in Barcelona. I'm doing a number of sessions at both events, but what's cool is that at each event I'm doing a talk specifically about CSLA .NET for Silverlight. And in December I'll be at VS Live in Dallas, where I'll also give a talk directly on using CSLA .NET.