Question

I have a table which I do a self-join. My sample table looks like this

-- Table genealogy
+--------+---------+------+-----+---------+-------+
| Field  | Type    | Null | Key | Default | Extra |
+--------+---------+------+-----+---------+-------+
| parent | int(11) | NO   | PRI | NULL    |       |
| child  | int(11) | NO   | PRI | NULL    |       |
+--------+---------+------+-----+---------+-------+
2 rows in set (0.01 sec)

The concept goes like this (get ready for this one):

1. To find the child of user parent=1, I look at the first level of user parent=1.
2. To find the grandchild of user parent=1, I look for the child who is the child of the child of user parent=1.
3. To get the great-grandchild (3rd level), I look for the child who is the child who is the child of the child of parent=1.
4. …and so forth

It can get a little confusing but my query works just fine. I was wondering if there’s a much faster/optimized way of doing it since I’m only repeating my sql all over the place.

-- Get Child
SELECT parent, child AS '1st Level' FROM genealogy WHERE parent=1;

-- Get Grandchild
SELECT a.parent, b.child AS '2nd Level' FROM (SELECT * FROM genealogy WHERE parent=1) a INNER JOIN genealogy b ON a.child=b.parent;

-- Get Great-Grandchild
SELECT a.parent, b.child AS '3rd Level' FROM (SELECT a.parent, b.child FROM (SELECT * FROM genealogy WHERE parent=1) a INNER JOIN genealogy b ON a.child=b.parent) a INNER JOIN genealogy b ON a.child=b.parent;

-- And the list goes on

Answer 1

You could use another tree structure (called nested set):

-- Table genealogy
+--------+---------+------+-----+---------+-------+
| Field  | Type    | Null | Key | Default | Extra |
+--------+---------+------+-----+---------+-------+
| id     | int(11) | NO   | PRI | NULL    |       |
| lft    | int(11) | NO   | NO  | NULL    |       |
| rgt    | int(11) | NO   | NO  | NULL    |       |
+--------+---------+------+-----+---------+-------+

This would enable you to select ALL childs (and grand childs and grand-grand-childs and so on) of a node inside the tree.

For example to select all children of genealogy id 1:

SELECT g2.*
FROM genealogy AS g1, genealogy AS g2
WHERE 
  (g1.lft BETWEEN g2.lft AND g2.rgt)
AND 
  (g1.id = 1);

You may have a look at this for more information and more examples: http://www.ibase.ru/devinfo/DBMSTrees/sqltrees.html