First, a data-oriented language that deals specificly with manipulating large datasets. This would basically work like LINQ, dynamically creating strongly typed classes for access to any database connection. Yes, Typed Datasets already do this; however, they
would be dynamic and always automatically reflect the data format that is in the database.
The primary purpose of the framework would be to manipulate data that is in XML or SQL Datasources beyond what any SQL query could do without needing to understand Stored Procedures. In the enviornment you would specific one or more various datasources. The
schema's of the datasources would then be reflected in the .net framework so IntelliSense would work when you're writing the data scripts. Finally, you could run the scripts and manipulate whatever datasource you want to manipulate. The scripts would be compiled
The second framework would be something like Ruby on Rails for the .net framework... another "Convention over Configuration" framework. If you haven't used Rails yet this being forced to build your application a certain way is really nice. It seems like its
very limiting at first but it actually makes developing easier, you don't have to think too much about how to build/structure the application correctly. The ASP.net web projects are more difficult to code MVC style; you have to think more about it anyway.