In my last mention of Typed DataSets, I commented on the overall fragility of the structure and I came out in support of using SP and data readers as a superior method of data management. I did state that the serialization that typed datasets gives you is one great reason to use them, but suggested that they work best with a data structure that isn't going to change a lot. I also feel they have a lot of overhead that the reader/SP approach doesn't.
Well, I've seen another good reason to use typed datasets despite the drawbacks: relations. I can get at child table data using the relationships established in my database, and do some on the fly filtering of the data at the client. This didn't immediately come to mind when I started using them, but now I see how powerful (and fast in the right context) this can be.
Now, admittedly this is cursor-based, and therefore perhaps not as fast a reader in some situations, but it is a nice, OO way to deal with your data.
I might become a fan of them yet.....
posted @ Tuesday, September 11, 2007 10:50 AM