We have a SQL Server database that we want to mirror in Oracle so we can use our LLBLGen code regardless of the backend database type. Obvious problems are column types in SQL Server are not supported in Oracle such as bit and guid. But not-so-obvious problems are limits on the database name size, RI name length restrictions, and table name length restrictions. So, in order for LLBLGen Pro to work in this case, do all of the database objects have to match in both names and types 100%?

What is the best way to approach this problem so we can design the database schema in a way that will work with either Oracle or SQL Server using the adapter approach? Any suggestions are greatly appreciated.

do all of the database objects have to match in both names and types 100%?

No, But Generated Enteties Classes along with their fields names and types should match. Entities & Fields names could be edited And to have the same generated fields(properties) types, you should use TypeConverters

Please read the section "Concepts - Type Converters" in the LLBLGen Pro documentation

Now you will have a common database generic project generated, and 2 database specific projects that you should use for each database.