On Thu, Jan 22, 2004 at 11:24:59PM -0500, Jeff Urlwin wrote: > > > > The "Driver" test would, IMHO, be VERY dependant upon the > > conformance > > > tests, to know what to test and to know how to approach inserts, > > > deletes, etc. > > > > What do you mean by "how to approach inserts ..."? > > Actually, it's more how to approach creating the temp table (getting the > DBMS specific sql type name for a given type), so that you *can* insert.
Right, and that's what I've said the DBI::Test per-dbtype subclass will do. > > I expect your DBD::ODBC tests will provide valuable input to > > Tom's design. > > I hope so. If nothing else, to see some of the things that I had to deal > with. Some drivers don't report support for SQL_INTEGER, but do support > SQL_SMALLINT because their integers do not reach the range of the ODBC spec. > I suspect that since it seems that some people are using DBD::ODBC to get > type info from their DB (thanks Steffan!), I suspect that issue will be > perpetuated. It would be interesting to include tests for the edge cases (min and max values etc) for some types like ints and floats. We could validate that what type_info returns is a) true as far as it goes (ie SQL_SMALLINT) and b) could be changed if the range of values/precision/scale etc is wide enough. But that's icing the cake we haven't baked yet. Tim.