So coming home from PDC turned out to be much different than expected…my wife has been bedridden with a cold since I stepped back into the house on Friday night. I was hoping to reflect a lot more on PDC over the weekend, but alas that was not to be (although it was nice to spend so much time with my daughter after being gone for a week.)
I got a lot out of PDC, but like a lot of things, it may not have been what was directly intended to be delivered, but rather what was between the lines that mattered. It was hard to get a solid track for session attendance – I tended to be all over the place…I think my next conference will have me going to a very narrow-focused set of sessions and then catching the videos for what I missed. Regardless, having the videos available is handy, and I’ve already watched several for sessions that I had to skip for one reason or another.
From a high-level, my thoughts are as follows:
Azure: Raymond Chen once blogged about the true measure of when a project is “real” being when stakeholders start talking more about what it won’t do than about what it will do. Azure seems to be there. A lot of the general-high-level functionality is in place, and they’ve managed to plugged some significant holes in very short-order (eg. single sign-on…) “Dallas” is big, and I think that once I am able to cobble together some demos, some people I know will find it irresistible. The general place where Azure lives and/or will live is in the ability to scale ASP.Net applications up to Azure (an interesting idea is to keep existing data centers, but use Azure for redundancy and for elastic scaling…) and soon to be able to revert Azure applications back down to ASP.Net and private data centers.
Silverlight: With the enhanced LOB features in SL4 (printing, right-click context menus, shared assemblies, etc.) and especially with the ability to run standalone SL with enhanced trust, the line between SL and WPF is getting incredibly blurry. Bottom line, Silverlight continues to be a platform worthy of the time spent becoming familiar with it.
Parallelism: The content here was not really new – Moore’s law seems to still be predicting 80 core machines in the not-too-distant future. The new .Net 4 support for parallelism builds nicely on top of what is already in the framework, but unfortunately there’s still nothing to replace the requisite “InvokeRequired” boilerplate checks at the presentation layer. This becomes problematic when a layer outside of the presentation layer gets refactored to use threading…the UI layer isn’t written expecting it, resulting in a runtime exception. Ideas involving an application-level attribute or other high-level approach to baking the thread marshaling code into the UI framework controls themselves would probably go a long way, and conversely, the absence thereof will probably stifle the extensive use of parallelism in real-world applications, due to the perceived complications that its inclusion introduce.
Data: Put simply – goodbye LINQ-to-SQL. With EF4, there’s really not much need anymore. The ability to come at data from either model-first, code-first, or database first is really helpful…from my perspective, when doing “Greenfield development“, the Model is My Truth, and both the code and the database are simply implementation artifacts.
What was missing: Ray Ozzie’s talk of “Three Screens and the Cloud” began to ring hollow when it became clear that Windows Mobile was taking a back-seat at this conference…no talk of WM7, Silverlight Mobile, etc. What happened to the “little screen?” Just saying “we don’t have it yet, but we’re working on it and just wait…it’s going to be awesome” would have gone a long way. Not mentioning it actually turned it into the 800-lb gorilla in the room.
Also, what happened to the Live Services story? What about Live Mesh? It looks like these topics are being taken back into the garage for a retune, and their inclusion in last year’s introduction of the Azure stack may have been inadvertent noise. I have found Mesh in particular to be a very useful tool, but its lack of any relationship whatsoever to SkyDrive is perplexing.
Finally, if this year is any indication (and it may not be), it looks like the PDC may be being positioned for a new identity. With other conferences like Mix, SQL-PASS, and SharePoint-specific conferences, among others, it may be time to make TechEd IT-specific and bring TechEd’s developer content into PDC. I felt the “split approach” taken by TechEd in 2008 (1 week for dev, 1 week for IT) worked out nicely. Time will tell…
As I said, it was a good conference for many reasons for me. In addition to the show contents, there were interesting networking experiences. I’ll be posting about individual technologies in the coming few days.