UNION without CORRESPONDING - SQL

The use of UNION without CORRESPONDING. Example is merely by omitting CORRESPONDING, but only because the operands have identical SELECT clauses.

Example: UNION without CORRESPONDING

SELECT StudentId

FROM IS_CALLED

WHERE Name = 'Devinder'

UNION DISTINCT

SELECT StudentId

FROM IS_ENROLLED_ON

WHERE CourseId = 'C1'

When CORRESPONDING is omitted, names are not used at all in the pairing of columns. Instead, SQL's definition, in yet another departure from relational database theory, depends on an ordering of the columns: the first column of the first operand is paired with the first column of the second operand, the second with the second, and so on. As with CORRESPONDING, columns thus paired do not have to be of the same type. Furthermore, the two operand tables must have the same number of columns, so that there is no unpaired column in either operand, also as in relational union.

Although the operand columns in still have the same name, StudentId, that is not a requirement in this variety of UNION. For example, SELECT StudentId AS X could be the SELECT clause of the second operand. However, if corresponding columns do not have the same name, then the corresponding column in the result is effectively anonymous (the standard defines it to have an unpredictable system- generated name). Actually, some implementations use the column names of the first operand here, thus destroying the normal commutativity of UNION. The user of an implementation that strictly follows the standard would perhaps be well advised always to make sure the corresponding columns have the same name anyway, to avoid the unpredictability of system-generated names and to improve portability from one implementation to another.