Hi,

While I think your heart is in the right place, I don't see this as very practical to implement for all database systems. The reason being is that some commercial DBMS vendors have literally thousands of possible errors with some being 'numbered' and others not. Also, depending on the version of the DBMS, a single error message might mean two completely different things.

If anyone was thinking of creating such a database of error messages, I would recommend to concentrate on the free and open source database systems first. Postgres, MySQL, SQL Lite, etc. If it becomes feasible, expanding it out to the commercial DBMS systems.

The real problem isn't so much the error messages but more of the context of the errors with respect to the application. I *do* like the idea though I suspect it would be better off explaning them at the application level instead. (duplicate key = dependents for Customer # 1883 already submitted. Would you like to resubmit them?)