> As I said - I prefer to have compiler errors instead of runtime errors > :-)
That is probably the only really compelling argument for typed datasets, although the intellisense is pretty nice too. I think this thread's concentration on an OR system really proves the point Ben and I have been saying though. We all understand a dataset is not a OR system. You could use a dataset in creating an OR and most people would benefit from doing so if their needs are simple. Everything discussed here probably falls in that last 10% of use cases where it simply doesn't make sense to force a dataset to do what it was not inteneded to do. What is unique about the dataset and reflects it's heritage in the XML world rather than the OO world, is that a UI can also understand it. If your OO logic implements the Memento pattern, you can fully retain encapsulation, and still allow the use of datasets in your OR approach, however you decide to do it. On the front end however, you get large productivity gains and still fully preserve the ability to allow your business objects to enforce rules, trigger various processes etc.. One other thing I like about this approach is that writing tests is so much easier. Just create your instances with data retrieved from a file in a dataset rather than from a db in a dataset. This sourcre agnostic ability also plays into the as yet imaginary world where we allow business partners to use our systems, possibly not even touching our database through web services. __________________________________________________ Do you Yahoo!? Faith Hill - Exclusive Performances, Videos & More http://faith.yahoo.com You can read messages from the Advanced DOTNET archive, unsubscribe from Advanced DOTNET, or subscribe to other DevelopMentor lists at http://discuss.develop.com.