Re: Implementing complicated constraints

"Tony Andrews" <andrewst_at_onetel.com> wrote in message news:1096546603.206373.64250_at_h37g2000oda.googlegroups.com...
| Mark C. Stock wrote:
| > however, since the ATXT can apply to only one AID, and ATXT can only
| exist
| > in the context of AID, id' remove TID from table A and change table T
| to:
| >
| > T(TID, CID ref C.CID, ATXT, C.AID) with unique constraint on CID,
| ATXT and
| > NN an UK on AID
|
| That still leaves unresolved the rule that T.CID should match B.CID for
| the same AID value.
|
| It is also not normalised, since T.CID is transitively dependent on
| T.AID.
|

started looking at the BID issue -- decided that we're shooting at phantom targets without more info from the OP(or rather, i've got to get back to real work and don't want to analyze the OP's statement of problem any further right now)

but you're right on the T.CID/T.AID issue -- my approach would still require CID in A, with a UK on (CID,AID) and an FK to it in T -- not sure if that's the way to go, but as i said, it's time to get back to real work

good discussion, though

++ mcs Received on Thu Sep 30 2004 - 07:42:36 CDT