Question

I have two mySQL tables in my database

Training:

+-------------+------------+------------+------------+-...-+------------+
| Training_ID | 09-06-2012 | 16-06-2012 | 23-06-2012 | ... | 28-12-2013 |
+-------------+------------+------------+------------+-...-+------------+
|     1       |    123     |    924     |   367      |     |   679      |
+-------------+------------+------------+------------+-...-+------------+
|     2       |    243     |    626     |   381      |     |   771      |
+-------------+------------+------------+------------+-...-+------------+
|     3       |    766     |    826     |   956      |     |   725      |
+-------------+------------+------------+------------+-...-+------------+

Users:

+---------+----------+
| User_ID | Name     |
+---------+----------+
|   123   | Fred     | 
+---------+----------+
|   924   | James    | 
+---------+----------+ 
|   367   | Tim      | 
+---------+----------+
|   766   | Mark     | 
+---------+----------+

What I want to show is:

+-------------+------------+-------------+------------+-...-+------------+
| Training_ID | 09-06-2012 | 16-06-2012  | 23-06-2012 | ... | 28-12-2013 |
+-------------+------------+-------------+------------+-...-+------------+
|     1       | 123 - Fred | 924 - James | 367 - Tim  |     | 679 - Ben  |
+-------------+------------+-------------+------------+-...-+------------+
|     2       | 243 - Sarah| 626 - Smith | 381 - Exam |     | 771 - John |
+-------------+------------+-------------+------------+-...-+------------+
|     3       | 766 - Mark | 826 - Jone  | 956 - Case |     | 725 - Brett|
+-------------+------------+-------------+------------+-...-+------------+

Note: there are actually 70 columns (all the same format as the above).

I tried using joins – but I cant do one for each column.

Is there a way to apply a join to the whole table?

Answer 1

That looks like a terrible design and any query you’re going to perform to produce your desired format will also look equally as horrid, but it isn’t impossible:

What you can do is make a single Cartesian-join of the training and users table, then GROUP BY the training_id, and make a bunch of conditional aggregations which will put everything on one row per training_id:

SELECT
    a.training_id,
    MAX(CASE WHEN a.`09-06-2012` = b.user_id THEN CONCAT(a.`09-06-2012`, ' - ', b.name) END) AS `09-06-2012`,
    MAX(CASE WHEN a.`16-06-2012` = b.user_id THEN CONCAT(a.`16-06-2012`, ' - ', b.name) END) AS `16-06-2012`,
    MAX(CASE WHEN a.`23-06-2012` = b.user_id THEN CONCAT(a.`23-06-2012`, ' - ', b.name) END) AS `23-06-2012`,
    MAX(CASE WHEN a.`28-12-2013` = b.user_id THEN CONCAT(a.`28-12-2013`, ' - ', b.name) END) AS `28-12-2013`
FROM training a
CROSS JOIN users b
GROUP BY a.training_id

You’ll need as many MAX(CASE WHEN… columns in your SELECT clause as there are columns in your table.

While this may also look ugly, it is much more efficient than joining the users table 70 different times for each column.

---

SQLFiddle Demo with just four columns