Tips for using Visual Studio Typed DataSets?

Question

When using strongly typed dataSets in Visual Studio 2005/2008, if the underlying database schema changes, the only practical way to refresh is to delete the dataset and recreate it from scratch. This is OK unless I need to customize the dataset.

Customizing by extending the partial dataset class allows customizations to be retained, but then a simple FillBy() again becomes a long sequence of SQL.

Is there any way to resynchronize a dataset with the database schema without losing dataset customizations?

Answer 1

If you are just making a simple change like adding a field to a table, then I right click on the table and click configure. Go through the wizard again adding the new field, (or even replacing the select with select *) and it adds the field to the table and syncs up all the queries that you've built. I've had tables with about 10 different custom queries and when I reconfigure the table the queries all updated properly.

Answer 2

Because typed datasets use generated nested classes, customizations will often be lost. What I do is generate the typed dataset, and then mark several classes as partial. Then I create separate files containing a duplicate nested class structure (also marked partial).

This way I can regenerate my data set and the only update I have to do is to go back and make them partial again. My customizations are kept in separate files.

Answer 3

Typed Datasets are evil. LINQ to SQL is what typed datasets should have been, so I like to think of it as Typed Datasets 3.0. I hope when they fold L2S into the Entity Framework they retain what makes L2S so good.

Answer 4

I've been using typed datasets quite successfully in a number of web and windows projects. There was a bit of discovery in the beginning - learning all the quirks of the dataset designer and how to extend the dataset to provide more flexible connection configurations. But once you get past that, generating a data access layer will be quick work.