Shannon Braun's Weblog

• Simple and straight forward, little configuration. 
• Post only functionality (I can refine the posts using the web UI if I need to, and will possibly add addtional functionality later)
• Acts like Microsoft Word (spell check, formatting, etc) when composing
• Save posts off to file system, so I can evolve it over multiple revisions if needed, or I don't have a readily available internet connection.
• It must use the HTTP protocol for the post submission (that way I can post reliably from anywhere with an internet connection)

SAF day 3 was much more of the same... Keynotes and discussion groups.  It was great for networking, but from a content stand point, it was OK.  I had seen a lot of the content through the APB(Architecture Partner Board) and the discussion groups varied based on who you had in the group.  Some were very good quality while other were not impressive at all.

The day didn't start to well... on my commute from my hotel to the Microsoft campus someone hit the rental car I was driving, so I spent the majority of the morning taking care of that.  I missed the morning keynote that dug more into SOA and was basically the second half of the Pat Helland's presentation.  Luckily I had seen most of that back in August when he was working on the initial draft.

I arrived at the conference center just before an informative session on writing client applications on the Longhorn platform using some of the new stacks (Avalon, Indigo, WinFS).  Although this is a long way off, there was some pretty neat UI demonstrated.  I was also interested in the fact that Longhorn will be fully managed and allow for complete access to the OS.  The advantages that we see now in the unified BCL are the same that we will see when the OS becomes managed.

OK... I didn't get a picture... but one of my co-workers at BORN (Ed Ferron) got a picture with Bill when he left the Q&A.  Nice job Ed... I knew I should have followed you.

SAF began with a welcome by Adam Denning, Director, .NET Architecture Team.  He discussed an overview of what the SAF is about, Microsoft sharing information and sharing ideas about architecture.  The initial key note was by Pat Helland... one of Microsoft's key “think tank” members.  He presented a 90 slide presentation on an analysis between how city/government and transportation progression (deemed Metropolis) of the last 200 years mirrors the evolution of technology now.  I had actually seen the presentation back in August at the APB (Architecture Partner Board) when it was about 8 hours and over 200 slides. This presentation was also shown at PDC, and you can get the slides here.

That keynote was followed by a Q&A that involved Bill Gates and Eric Rudder. 

I have been working with the XmlSerialization stack on and off for about 2 years now. My first introduction to it was at DevelopMentor's Guerilla .NET @ Microsoft course back in November 2001. I had used it for trivial serialization tasks and as a simple topic for presentations that I gave internally for other consultants and at Microsoft DevCare. It was a straight forward, simple way to serialize object graphs to a human readable format (Xml).

Though useful, it was rigid and didn't always meet my requirements. In those instances I took advantage of the more robust System.Runtime.Serialization stack. After all, that stack supported private/public field serialization, highly customizable output through custom formatters and had a higher type fidelity. In addition, I could supply serialization for types that were not marked [Serializable] (through the ISerializationSurrogate).

It wasn't until the last couple of months that I really discovered that the System.Xml.Serialization stack was more flexible then I had originally thought. I stumbled across the flexibility when I was trying to emulate the ISerializable interface found in the System.Runtime.Serialization stack for an extended XmlSerializer class I was implementing to address some the original stack's short comings. I had tried implementing ISerializable on my class, then serializing it using the XmlSerializer, but none of the interface methods were called. By chance I created a IXmlSerializable interface... when I compiled, it failed indicating that the interface already existed. When I read the API docks on it, it was exactly what I was trying to implement. Prior to that I thought the System.Xml.Serialization stack was limited to a number of attributes (XmlAttribute, XmlElementAttribute, XmlIgnore, XmlArray, etc) used to describe how/what you wanted the XmlSerializer to serialize in your object graph.

What I discovered through the research is that there was more flexibility with the System.Xml.XmlSerializer then initially thought. By implementing IXmlSerializable in my classes, I could not only control how I formatted my Xml, but I could also serialize private members. Though it is a bit more work, because you must supply the implementation for not only writing the object to an XmlWriter, but also supply an implementation using the XmlReader to hydrate your object. IXmlSerializable is the equivalent to ISerializable in the System.Runtime.Serialization stack.

Granted this still doesn't give you the power that the System.Runtime.Serialization stack supplies, but it does open the door for additional implementations that may have been overlooked in the past.

Just finished packing for my trip to Seattle.  The SAF (Strategic Architect Forum) for Partners starts tomorrow, so I am taking the 6:30am out of Minneapolis to SeaTac.  I like getting back to Seattle (after living there for 5 years), it feels familiar.  It also gives me a chance to hook up with friends from there.  Although this trip is pretty short, I am coming home Saturday morning, not staying the weekend like I did last time.

 Here are some of the session that I plan on attending while I am there:

SQL Server Next: Yukon 
Discuss the features and uses of the "Yukon" release of SQL Server. How will hosting the CLR in SQL enhance SQL programmability? How can the SQL service broker be used to deliver web services from the database? Are there pitfalls, and how can they be avoided?

Architecting for Real-time Business Intelligence 
The opportunity to profit from information is inversely proportional to the time it takes you to act on that information. How can we model applications and service portfolios to mine key data on the fly to discover interesting correlations?

Smart Client
What role should smart clients play in your application portfolio? How will the advances in the "Longhorn" programming model make smart clients more compelling? What is the migration path from Win32 to .NET to Longhorn?

Portal Solutions
A look at he good, bad, and ugly of portals -- from business-to-employee solutions to professional service automation portals. What makes good solutions succeed and bad ones fail? How can correlation and coordination across portal "parts" be used to improve the value of the solution? Where is Microsoft's portal platform going, and where should it go after that?

EAI and Integration Architecture
How can we model our existing application portfolio to discover opportunities for integration? How do we act on that model, using service facades, connectors, and legacy-integration technologies?

I will try to blog while I'm there but the schedule looks pretty packed. 

Download: AppBlocksWebCast-FINAL.zip