Hi,
I'd like to use a lookup-table that contains text-values and their keys.
There is a data-table with entities that have (among others) 2
attributes which could nicely be represented by numbers.
e.g. enttity (e_id, ..., color_fk, shape_fk, ...)
Now I could create 2 tables with the text:
color (-1, 'not provided'), (0, NULL), (1, 'red'), (2, 'green'), (3,
'blue'), ...
shape (-1, 'not provided'), (0, NULL), (1, 'point'), (2, 'line'), (3,
'triangle'), (4, 'circle'), ...
There might come more such attributes.
Therfore I'd rather have just 1 attribute-table with a 3rd column that
has a group-id:
attributes (attribute_id, attribute, attribute_group)
(-1, 'not provided', 0), (0, NULL, 0),
(100, 'red', 1), (110, 'green', 1), (120, 'blue', 1), ...
(200, 'point', 2), (210, 'line', 2), (220, 'triangle', 2), (230,
'circle', 2), ...
Easy:
Within the entities-table I can define 2 fk-constraints on color_fk,
shape_fk which reference the cumulative attribute-table.
The application should make sure that the user could only chose from
group 0 (shared) and one other.
Difficult
I'd like to let the DB control this to not risk that shape-ids end up in
the color_fk.
How can I make shure, that color_fk only receives values from
attribute-groups 0 and 1 but no other?
Probaply I got it philosophically wrong ?
