Question

Let’s say I have a table called ‘species’ with 3 columns: ‘Id’, ‘ancestorId’ and ‘name’. The ‘ancestorId’ is the ‘Id’ of the ancestor specie, so for example if the ancestor of the homo sapiens is the australopithecus, and australopithecus’s ‘Id’ is 5, then homo sapiens’s ‘ancestorId’ is 5. Let’s say, also, that there is a specie called ‘First specie’ whose ‘Id’ is either 0 or null. So what I want is to select a list of ancestors from a particular specie, say, homo sapiens, to the ‘First specie’, no matter how many nodes are on the way. Is this possible with SQL?

Answer 1

The ANSI way is to use a recursive WITH clause:

WITH hierarchy AS (
  SELECT t.id, 
         t.name,
         t.ancestor
    FROM TABLE t
   WHERE t.ancestor IS NULL
  UNION
  SELECT t.id, 
         t.name,
         t.ancestor
    FROM TABLE t
    JOIN hierarchy h ON h.ancestorid = t.id)
SELECT *
  FROM hierarchy

Supported by:

• SQL Server 2005+
• Oracle 11gR2
• PostgreSQL 8.4+

Oracle’s had hierarchical query support since v2, using CONNECT BY syntax.