SQL: Table, where "partnerships" are stored

[softwarejaeger]

Hello, i have a little problem. I have a table "customers" and a table "partnerships". In the customer database i have stored a few customer with a ID as primary key. In my partnership database i have two columns (partner1, partner2) where i want to store partnerships between them. So now the problem, it should be unnessecary if partner1=1 and partner2=2 or partner1=2 and partner2=1. And one partnership shouldn't be stored two times. So how can i realize this? In my selects it is easy, i easily have: select * from partnerships where (partner1='1' and partner2='2') or (partner1='2' and partner2='1') But how to make the right constraints, that this isn't saved to times, so i don't want to have something like this:

partner1 partner2
1 2
2 1

Maybe someone has a solution for me?

[Mycroft Holmes]

I cant see a simple way to do this using relationships, you could set up a complex structure that concatenates the IDs in order of size and faff about with that but it would be ugly as sin. I'd just have a proc that checked for existence before inserting. OR make it a convention that the highest is always Parent 1, this does not help with constraints though it would simplify your query.

Never underestimate the power of human stupidity RAH