>>My understanding is that Ideablade creates a "link file" or a shortcut for a given class object from the Silverlight side to the web side through the Ideablade template that is generally used in Ideablade applications. When the project compiles, Ideablade directs the compiler to the same physical file, which resides on the web side of the project, thus making sure that both Silverlight and Web sides compile exactly the same code. RIA, on the other hand, creates a proxy on the SL side, which has its own code that is used when compiling the SL side of the project. Since this proxy file is physically different from the web side file, these two files can get out of synch at which point the project can spin out of control.
>>
>>If you are further interested in this (and actually hearing it straight from the "horse's mouth", I suggest watching this video (this particular explanation starts around minute 3): http://www.ideablade.com/Videos/SilverlightQuickie/index.html
>
>Thx, I'd be interested in looking at the Ideablade take on this. I haven't watched the video yet but it sounds as if the MS 'Silverlight Business Application' template does pretty much the same thing for some classes (you can do it yourself in the IDE by choosing 'Add as link' when adding an existing item)
>
>But the problem there is that often the class on the server side may have a lot of dependencies based on libraries that are not part of the Silverlight package so there isn't really an alternative to building a 'proxy' version for the client side.
>
>In my limited experience RIA services does a reasonable job of generating the client-side code for this when the server side changes. What doesn't work so well (if at all, using the Entity Framework) is the synchronizing of the EF model with the DomainService. If you need to make a change to the edmx then the DomainService has to be manually destroyed and completely recreated. And since the DomainService is not created as a partial class then any additions/changes have to be completely redone......

I am not pushing Ideablade on anyone, by the way, but I have found it very useful in my own work because it takes care of this and some other holes in the SL and/or RIA way of doing things that I would otherwise have to ever so carefully code, manage and maintain manually. Ideablade is not perfect, but it does help me a lot. And given that they have a free version with no super serious limitations as well as a boatload of great walk-through videos and sample projects and code, it really is pretty easy to get started with.

It still puzzles me, though, how cumbersome data handling in .NET continues to be compared to some ancient tried and true technologies out there. It is getting better, granted, but IMO it is still much harder than it *should* be. And then there is the whole painful process of changing backend datamodels and having the changes propagate through a complex system. I would like to use a simple Case tool (like xCase) to manage my schema(s), and with a click of a button refresh my application with all changes on the schema AND the backend. Is that too much to ask? It appears that without a data dictionary/metadata -driven approach it actually IS too much to ask. Maybe I've just had it too good with some of the ancient technologies I've used successfully over the years...