Another big learning day on Day 4.  I started with Michele Leroux Bustamante's "Patterns for ClickOnce Deployment and Versioning."  ClickOnce always seems so cool, and I learned things I hadn't from a few other prior demos.  I like the option where you can require the user to download the new version before running the application again.  This is especially useful in an enterprise environment in situations such as breaking changes to a database schema.  Keep in mind that this option nullifies the "revert to prior version" option when uninstalling, for obvious reasons.

There's a question I asked which Michele didn't know the answer for (she considered it a "blonde moment" -- her words, not mine -- she is very funny), so I have to try it out, myself.  You can publish an update that only updates selected components.  So my question was, if a user accidentally deletes required components of an app (even the EXE), would the updating technology automatically re-download the missing components, or only the newly published components.  I have a hunch we'd need a separate "full install" option that the user can use to back-fill.

Michelle is one of Juval Lowe's partners, and despite that fact they are both very funny (in completely different ways), they are as different personality-wise as you can imagine.  I'm looking forward to taking an architecture course from their company, IDesign, later this year.

The next session was Rocky Lhotka's "Disconnected Windows Forms Application Architecture."  I wanted to take this session to learn some alternative approaches to asynchronous processing; each strategy has its obvious and not-so-obvious applications.  This was definitely interesting.  I was considering MSMQ as an approach, and was also going to look into Microsoft's code block for this, but now I'm having second thoughts.  Web Services seems like a better approach, but I want to read up on this in more detail before taking on this challenge.  I also have to check out Rocky's updated book on his framework.  I've heard a lot of great things on .NET Rocks! about it.

Next came .NET Rocks! Live!, with guest Kathleen Dollard.  And like I mentioned after her session, she definitely rocks!  We had a relatively small crowd for such a hugely popular show, but when I reviewed the list of speakers for the same time period (Dino Esposito, Brian Noyes, Juval Lowe, Dan Appleman), it made total sense.  The scheduling gods screwed this one up.

But I just had to go to this session, considering that the show has had a huge impact on me over the past few years.  It was great finally meeting Richard Campbell.  Now I loved Rory Blyth as a host (he's bleeping insane), and Mark Dunn was great, also, but Richard is perfect, in my opinion.  He fits perfectly with Carl Franklin, his questions are on target, and he has a tremendous knack for being extremely explicit and clear in his questions and explanations.  I think the show has risen to a new level since he became co-host.

This episode was mainly to "catch up" with Kathleen, about what she was thinking about lately in the industry, code generation (her specialty), TDD, and several other topics.  This show flowed great (as everyone will hear soon when the download is available).  I also made my second (although unannounced) "appearance" on this episode with a comment about TDD at the start of the second half (if they don't edit it out, of course).  My first time was on episode 67, when I was in the studio for that recording, and all of a sudden Carl wanted my and my former business partner's opinion on pair programming.  I won't go into much detail about this session, because you'll be able to download it shortly.

I finished the day out with Microsoft's Stephen Toub's session on "Multithreading Changes in the .NET Framework 2.0."  Multithreading is a fascinating subject, but it can be overwhelming.  One day, I'd like to post some entries here to try to explain it very clearly.  But I have a lot to learn myself before I "grok" it.  He did a fine job, but there was a hell of a lot to cover, and the best way to learn this topic is through experimentation.  But the addition of the BackgroundWorker capability in 2.0 is enormous.  If nothing else, you MUST read up on this.  There is no more excuse to take chances with UI processing across a worker thread.