dumping data from multiple tables with common fields - mysql

Question

Hey All.
I have a bunch of tables that have some common fields tying them together, but I can't figure out the right way to dump them in a meaningful way.
Basically, users will be given two tests, and each test may be taken multiple times.
The main table stores information about the user and the test, similar to the below (we'll call this table MAIN):
user_id test iteration completion_time
1 1 1 1:00
1 2 1 1:30
1 1 2 0:49
1 2 2 1:30
Each test page then has its own table to store the answers provided, since some pages have a ton of questions. We'll call this one sample table RESULTS, but there are many tables like this that are basically the same.
user_id test iteration q1 q2 q3
1 1 1 A B A
1 2 1 B B A
1 1 2 A B B
1 2 2 A B B
These results tables (again, there are many) basically store the results, plus just enough information to accurately tie the results together across all tables. I set it up this way because to use just one table for the results would have left me with a table with several hundred columns, which I had read was not recommended.
So the problem here is i can't figure out how best to tie together these tables and get the results out. I've read up on joins and unions and neither one seems right, as far as I can tell, because i need to pull data from ~10-15 tables at once.
I can do a huge complex select -- something along the lines of
select m.*, a.*, b.*, c.* from main m, results_a a, results_b b, results_c c where a.user_id=m.user_id and b.user_id=m.user_id and c.user_id=m.user_id'
and that works, but there's got to be a better way. Keep in mind that i've only given 3 results tables in this example -- in my actual application, it's going to be more like 15-20 tables of results.
Beyond being really complicated, it returns duplicates of some rows, and if i want to throw in any extra logic (lets say, for example, that i only want the same data i queried before, but only for test 2) it gets even more complicated. And lets not even talk about sorting.
From what I've read, JOIN is for 2 tables, and UNION combines rows of results, not columns.
I don't claim to be a mysql expert, but i have looked into this before posting. I feel like I must be close to the right answer, but just not quite hitting on it.
Can anyone give me some guidance?

Answer 1

To use inner joins try:
SELECT m.*, a.*, b.*, c.*
FROM main m
INNER JOIN results_a a ON a.user_id = m.user_id
INNER JOIN results_b b ON b.user_id = m.user_id
INNER JOIN results_c c ON c.user_id = m.user_id
WHERE m.user_id = x
To differentiate column names, explicitly name the column and assign an alias
SELECT m.*, a.q1 as A_Q1, a.q2 as A_Q2, b.q1 as B_Q1, b.q2 as B_Q2 ...

Answer 2

I think you've designed your database in a way that you're going to have to do joins. It sounds better than the alternative (large, non-normalized table). What are you trying to get out? View all tests and determine a user's best score?
So if I understand the tables you have multiple users (obviously). Each user can take multiple tests. Each test could be taken multiple times.
select m.*, a.*, b.*, c.* from main m
join results_a a
on a.user_id = m.user_id and a.test_id = m.test_id and a.iteration_id = m.iteration_id
join results_b b
on b.user_id = m.user_id and b.test_id = m.test_id and b.iteration_id = m.iteration_id
join results_c c
on c.user_id = m.user_id and c.test_id = m.test_id and c.iteration_id = m.iteration_id
That should give you one row per iteration, per test, per user. If you want a certain test for a user then add:
where m.user_id = #userid and m.test_id = #testid
and then you can look at all the iterations for a test, by a user.
If you want the latest iteration:
select top(1) ...
where m.user_id = #userid and m.test_id = #testid
order by m.iteration_id desc
You might consider generating a view when your questions are set-up so you don't have to generate the select statement at run-time which represents a particular test.