COmanage Developers List

[Benn Oshrin <>]

I think we'll probably want to discuss this on tomorrow's call, but 
here's a writeup of the options that have been proposed thus far:

[1] VARCHAR

This was my original proposal. Basically, each option gets a character 
representation, and each elected value gets encoded in the VARCHAR. For 
example, "race" has 5 possible values and so would be a VARCHAR(5). A 
person who identified as "asian" and "black" might be represented as "AB".

[2] INTEGER

Similar to VARCHAR, but using binary logic. Thus "asian" and "black" 
might be represented as 1|4 = 5.

[3] WIDE TABLE

Add one column per possible value. Thus

 INSERT INTO cm_demographics (co_person_id, race_asian, race_black)
 VALUES (123, true, true);

[4] MANY-TO-MANY

Instead of having one table, we have one table per demographic type. Thus

 INSERT INTO cm_demographics_race (co_person_id, value)
 VALUES (123, 'asian');
 INSERT INTO cm_demographics_race (co_person_id, value)
 VALUES (123, 'black');

In both cases [1] and [2], the model translates between the encoding and 
the view representation.

Note the cm_demographics wiki page is currently a hybrid of [1] and [2]. 
We'll need to fix that once we decide on a path.

Any other proposals to throw out there before we decide?

-Benn-