Our handling of precision/scale is inconsistent. There are assumptions being made in the JDBC client, types.dat, the parser, and other paths that need to be explicit and consistent.