The idea was only to have one column CID_LOCATION, so that
a) I can invent new _location_tables_ without changing table TTREES
b) I later can easy divide a _location_table_ into more tables a la
TFORRESTS -> TFORRESTS_COUNTY1/TFORREST_COUNTY2 ... and back
c) use a normal fk_constraint

c) doesn't work
b) too theoretical
a) seem to be more problems than advantages

To build subtables and supertables is more overhead than I want

So I do as I dis and if I understood correctly this was an option some
of you mentioned:

a) For each _location_table_ an ID- column
b) after-insert-trigger, that exact 1 ID is not NULL

Thank you all
Andreas Mosmann

--
wenn email, dann AndreasMosmann <bei> web <punkt> de