So PDC is over for this time, but PDC 2009 is already announced, so we might not need to wait three years to the next. In this post I will try to summaries a great week.

Windows Azure

I think this announment was expected, but I must say that the first impression is that it seems quite complete. After having the head spinning for a while I can see a fit for at least one of my customers. I guess that it will take a while, getting used to the architecture and other more business related issues (like policies, laws and so on), but cloud computing will continue to increase. Whithout being any expert on the offerings that Amazon, Salesforce and Google has, I think that Microsofts announced offering probably is the strongest and  most complete. This means that it (for now) seems that Microsoft is about to do it again. In the past they have managed to handle the "threats" from WordPerfect, Netscape, open source and many others and now they are hitting hard against Google with a complete strategi containing web, clients and phones.

Of course it's hard to make any final conclusions on Azure without any information of the prizing model, which of course is an important part of the overall value proposition.

.NET Services

The Service Bus has some really impressive features and the top one is the connectivty it enables between computers without opening any inbound ports (and also without opening outbound even if it will go a lot faster with some outbound ports opended). The bus has support for both direct point-to-point connections as well as publish/subscribe, but the pub/sub does not have queue support yet. The bus is protected through the Access Control Service which is a Security Token Service (STS), based on Geneva if I'm not misstaken. Microsoft have been working hard on interoperabiltiy and showed both java and Ruby demos using the advanced WS-* protocols involved when using a STS.
.NET Services will not be available (at least not known) for installation, and the main reason is that the services are built for massive scale and that means, today at least 50 nodes is required. For this CTP they've got their own infrastructure, but it will be moved to Azure.

REST is big now and everything Microsoft showed (Azure, .NET Services, Windows Live, and so on) all has REST api's. I think that it is in some ways a new Microsoft that showed itself during PDC, a company that are working hard on supporting standards. I got my own doubts about REST cleared, when I now have learned how it is possible to use together with STS without losing the simplicity of REST, but I will save that for an upcoming post of it's own.

Links: Clemens Vasters

Windows Live

Microsofts different Live offerings have been around for a while, but I have never had any feeling that it has been used broadly by developers. This will change now. I'm quite sure that I will use it in future projects, even though it is easier to come up with ideas that is useable for private users rather than coporate users. Mesh is one of the most likely services, but others can also be interesting. I can't put my finger on why I feel that Windows Live will get a break through right now. They announced Live Framework, which might be one reason, but the REST api's seems to be just as easy to use. Maybe it is just me that needed this push?

Links: http://dev.live.com/

"Oslo"

Even before the PDC it was known that the original "Oslo" code name now only included the modelling part and other things have been pushed to for example .NET 4 and "Dublin". Still, my impression of Oslo is that it was not even half baked and the demos they did was terrible, but first thing first. Oslo is three things:

• A language called "M", that is used to create textual domain specific langugaes (DSL).
• A visual tool "Quadrant", for displaying things in the repository and to create visual domain specific diagrams.
• A "Repository" for managing domain models.

The new M language is easiest of the three to grasp and look like a nice way to create textual DSL's. So far so good. For the other two I must say that I still feel very doubtful. The demos in the breakout sessions was terrible and missleading and I have never felt so dumb after a session as after one of the Oslo sessions. They used for example a Music Library as an example that they modeled and stored in the repository, and for me this is something that would never be stored in a repository (if not making a repository equal to a database). I (together with my colleagues) had a long chat with one in the Oslo team about the repository and without his clarifications I would have had no idea of what to think about the repository part. He did confirm though that it was a bad example that was done because many developers find "processes so boring". I think this was a very bad call and probably has confused a lot of this developers in positioning Oslo.

Even after the long discussion in "Ask the Experts" night I'm not sure on what Oslo will be when it's finished and I must get my hands dirty with it to make any real conclusions. There have been many tries before to make development model-driven, and it has been just as many failures. If "Oslo" will succedd is way to early to say, but I'm certain, that it at least will result in useable parts no matter what. Many well respected bloggers seems very exited though, and some of them are also suprisingly so for me.

Links: http://msdn.microsoft.com/oslo, http://modelsremixed.com/, Martin Fowler on Oslo, Sam Gentile on Oslo

Windows Presentation Foundation (WPF)

WPF is starting to take over completly. In fact I don't think that I saw a single windows form based application during this week. The applications does not have the flash and bling, much seen so far, but it is rather a way to bring a better and richer experience to the user and Visual Studio 2010 is just one of the examples. Silverlight is also making great progress and I think it is now time to start use this two technologies in the regular business applications. For me personally, the first abstacle is to find out how to do the basic organisation of the application as a whole and specially the navigation.

Links: http://windowsclient.net/wpf/

LINQ to SQL Is More or Less Dead

This is very obvious after the PDC week and Microsoft is open about it. All investment will be done in Entity Framework from now on. I have used LINQ to SQL in a recent project and the experience haven't been that good, even though this is a simple scenario. Fact is that LINQ to SQL misses to much to be useful to much more than demos. I guess (or hope) upgrading to EF won't be that hard.

Summary

Unfortunately I think there was to many sessions that did not reach my (high) expectations of technical depth, and there were also more speakers that did not perfom as well as 2005. Even though I still rank PDC as the most important and high quality Microsoft event and I would recommend everyone that has a chance to go there. I can promise that you will come back very inspired.