On Fri, Mar 8, 2013 at 1:26 PM, James K. Lowden <jklow...@schemamania.org> wrote: > On Thu, 7 Mar 2013 19:20:44 +0100 > Petite Abeille <petite.abei...@gmail.com> wrote: >> Yeah? 'cool' is not necessarily how I would describe it? having a >> check constraint 'magically' coerce - change! - the inserted data >> type is? well? not cool. I would call it a misfeature :) > > To be clear, the contraints Nico described don't coerce anything. By > preventing the insertion of values outside the column's domain (per its > affinity), the constraints ensure that future coersions -- as part of a > join, say -- won't lose data.
This. Thanks Jim. I dunno, I prefer strong typing, but I'm not going to dictate that it must be so. I like having options. Also, even a fan of strong typing will occasionally have handy uses for duck typing. So I see nothing terribly wrong here except with the way CAST to numeric works, and there's a simple workaround for that, so in the end, nothing terribly wrong. Nico -- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users