Some infos as requested:
I've finalized my own QSqlQuery subclass but in a way that is very specific to my needs. What I have:
3 Classes for a database repository that knows what the tables, fields and primary keys should be. Why? First, the database may be in a state that is corrupt. Second, the repository knows much more than the few infos that the database is able to give me.
My own query class derived from QSqlQuery. The class is able to generate SQL statements fro select, insert, update, delete. The select part is able to generate correct joins. Currently I work on support for outer joins on all supported databases.
My own delegates can access the repository and in a table view draw the columns in an application specific way.
Bookmarks