|
||||||||||||||||||||||
scalar-expression scalar-operator scalar-expression |
This document covers the following topics:
A scalar-expression
consists of
a factor or other scalar expressions including scalar operators.
Concerning reference priority, scalar expressions behave as follows:
When a non-qualified variable name is specified in a scalar expression, the first approach is to resolve the variable name as column name of the referenced table.
If no column with the specified name is available in the referenced table, Natural tries to resolve this variable as a Natural user-defined variable (host variable).
+
|
||
-
|
||
*
|
||
/
|
A scalar-operator
can be any of
the operators listed above; the minus (-) and slash (/) operators must be
separated by at least one blank from preceding operators.
A factor
can consist of one of
the items listed in the above diagram and described in the text below.
An atom can be either a
parameter
or a
constant
(as described in the section
Basic Syntactical Items).
column-name |
A column-reference
is a column
name optionally qualified by either a
table-name
or a
correlation-name
(see also the section
Basic Syntactical
Items). Qualified names are often clearer than unqualified
names and sometimes they are essential.
Note:
A table name in this context must not be qualified explicitly with
an authorization identifier. Use a correlation name instead if you need a
qualified table name.
If a column is referenced by a
table-name
or
correlation-name
, it must be contained
in the corresponding table. If neither a
table-name
nor a
correlation-name
is specified, the
respective column must be in one of the tables specified in the
FROM
clause (see Table
Expression).
COUNT
|
|||||||
( |
SQL provides a number of special functions to enhance its basic retrieval power. The so-called SQL aggregate functions currently available and supported by Natural are:
Apart from COUNT(*)
, each of these functions operates on
the collection of scalar values in an argument (that is, a single column or a
scalar-expression
) and produces a
scalar value as its result.
Example:
DEFINE DATA LOCAL 1 AVGAGE (I2) END-DEFINE ... SELECT AVG (AGE) INTO AVGAGE FROM SQL-PERSONNEL ...
In general, the argument can optionally be preceded by
the keyword DISTINCT
to eliminate redundant duplicate values
before the function is applied.
If DISTINCT
is specified, the argument must be the name of
a single column; if DISTINCT
is omitted, the argument can consist
of a general scalar-expression
.
DISTINCT
is not allowed with the
special function COUNT(*)
, which is provided to count all rows
without eliminating any duplicates.
USER
|
A reference to a
special-register
returns a scalar
value.