My understanding is when you reference another table you are referencing the hidden column autoid, which indexes its unique columns. So a reference to another table is always to its unique columns. Unique columns are only needed in the table being referenced to, which my example has (upper table references the unique 0 100 1 and 0 200 1 from middletable, which in turn reference the unique 0 100 and 0 200 from lowertable). The problem, as Devon found, is that JDB doesn't seem to handle references to unique data which is unique only by reference to another table (that is, if we instead had 0 100 0 and 0 200 1 in middletable we'd be fine, since the uniqueness is in middletable without its reference to lowertable).
Jordan -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Raul Miller Sent: Wednesday, September 23, 2009 11:59 AM To: Programming forum Subject: Re: [Jprogramming] JDB wrong data On Tue, Sep 22, 2009 at 11:52 PM, Devon McCormick <[email protected]> wrote: > Not sure I really understand this, but if "midkey" has unique values, JDB's documentation says it expects unique columns for references to other tables. -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
