http://developer.mimer.com

Data Types in SQL Statements

---

Mimer SQL supports the following data type categories:

• Character strings, see Character Strings
• National character strings, see National Character Strings
• Binary, see Binary
• Numeric, see Numerical
• Datetime, see Datetime
• Interval, see Interval.

In SQL statements, you make explicit data type references when creating tables and domains and altering tables. You also use data types in CAST and stored procedure variable declarations.

In addition, there is also a Mimer SQL ROW type that can be used in stored procedures, for more information see ROW Data Type.

Character Strings

The character string data type stores sequences of bytes that represent alphanumeric data.

The character string data type category contains the following data types:

Data Type	Abbreviation	Description	Range
CHARACTER(n)	CHAR(n)	Character string, fixed length n.	1 £ n £ 15 000
CHARACTER VARYING(n)	VARCHAR(n) CHAR VARYING(n)	Variable length character string, maximum length n.	1 £ n £ 15 000
CHARACTER LARGE OBJECT(n[K|M|G])	CLOB(n[K|M|G])	Variable length character string measured in characters.	For information on the object length, see Specifying the CLOB Length.

CHARACTER or CHAR

The CHARACTER (CHAR) data type stores string values of fixed length in a column.

You specify the length of the CHAR data type as the length of the column when you create a table. You can specify the length to be any value between 1 and 15 000.

When Mimer SQL stores values in a column defined as CHAR, it right-pads the values with spaces to conform with the specified column length.

Note: If you define a data type as CHARACTER or CHAR, that is, without specifying a length, Mimer SQL assumes that the length of the data type is 1.

CHARACTER VARYING or CHAR VARYING or VARCHAR

The CHARACTER VARYING, abbreviated CHAR VARYING or VARCHAR, data type stores strings of varying length.

You specify the maximum length of the VARCHAR data type as the length of the column when you create a table. You can specify the length to be between 1 and 15 000.

CHARACTER LARGE OBJECT or CLOB

The CHARACTER LARGE OBJECT (CLOB) data type stores character string values of varying length up to the maximum specified as the large object length (n[K|M|G]).

The large object length is n, optionally multiplied by K|M|G.

You can specify the maximum length of the CLOB data type as the length of the column when you create the table.

Specifying the CLOB Length

If you specify <n>K (kilobyte), the length is <n> multiplied by 1 024.

If you specify <n>M (megabyte), the length is <n> multiplied by 1 048 576.

If you specify <n>G (gigabyte), the length is <n> multiplied by 1 073 741 824.

If you do not specify large object length, the length is implementation-defined to 1Mb.

Maximum CLOB Length

The maximum length of a CLOB is determined by the amount of disk space available for its storage.

Using CLOBs

You can work with CLOBs as follows:

• Retrieving CLOBs with simple column references in the SELECT clause of a SELECT statement
• Assigning CLOBs using INSERT statements with a VALUES clause
• Assigning CLOBs using UPDATE statements
• Adding CLOB columns using CREATE TABLE or ALTER TABLE
• Dropping CLOB columns using ALTER TABLE
• Altering CLOB column data types using ALTER TABLE

There are some restrictions associated with using CLOBs. The only comparison supported for CLOB values is using the NULL predicate.

A CLOB column may not be part of any primary key, index, unique constraint or primary key constraint.

The comparison restrictions also prevent CLOB columns from being used in DISTINCT, GROUP BY and ORDER BY clauses and UNION statements.

You cannot: use CLOBs in stored procedures, define triggers for tables that contain CLOB columns or add CLOB columns to tables that have triggers.

About Character Strings

All character strings have a collation attribute. A collation determines the order for ordering and comparisons, see the Mimer SQL User's Manual, chapter 6, Collations, for a detailed description of collations.

National Character Strings

The national character string data type stores sequences of Unicode characters that represent alphanumeric data. Mimer SQL supports the Unicode 3.2 standard, see www.unicode.org/unicode/reports/tr28/ Unicode 3.2, for more information. The UTF32 encoding is used.

The national character string data type category contains the following data types:

Data Type	Abbreviation	Description	Range
NATIONAL CHARACTER(n)	NATIONAL CHAR(n) NCHAR(n)	National character string, fixed length n.	1 <= n <= 5 000
NATIONAL CHARACTER VARYING(n)	NATIONAL CHAR VARYING(n) NCHAR VARYING(n)	Variable length, national character string, maximum length n.	1 <= n <= 5 000

NATIONAL CHARACTER or NATIONAL CHAR or NCHAR

The NATIONAL CHARACTER (NCHAR) data type stores string values of fixed length in a column. You specify the length of the NATIONAL CHARACTER data type as the length of the column when you create a table. You can specify the length to be any value between 1 and 5 000.

When Mimer SQL stores values in a column defined as NATIONAL CHARACTER, it right-pads the values with spaces to conform with the specified column length.

Note: If you define a data type as NATIONAL CHARACTER or NCHAR, that is, without specifying a length, Mimer SQL assumes that the length of the data type is 1.

NATIONAL CHARACTER VARYING or NATIONAL CHAR VARYING or NCHAR VARYING

The NATIONAL CHARACTER VARYING, abbreviated NATIONAL CHAR VARYING or NCHAR VARYING, data type stores strings of varying length.

You specify the maximum length of the NATIONAL CHARACTER VARYING data type as the length of the column when you create a table. You can specify the length to be between 1 and 5 000.

Binary

The binary data type stores a sequence of bytes that does not represent alphanumeric characters.

The binary data type category contains the following data types:

Data Type	Abbreviation	Description	Range
BINARY(n)	N/A	Fixed length binary string, maximum length n.	1 £ n £ 15 000
BINARY VARYING(n)	VARBINARY(n)	Variable length binary string, maximum length n.	1 £ n £ 15 000
BINARY LARGE OBJECT(n[K|M|G])	BLOB(n[K|M|G])	Variable length binary string measured in octets.	For information on the object length, see Specifying the BLOB Length.

Note: In interactive SQL, how binary data is displayed depends on how the interactive SQL tool is configured. For example, in Mimer BSQL, binary data is displayed as its hexadecimal value.

BINARY LARGE OBJECT or BLOB

The BINARY LARGE OBJECT or BLOB data type stores binary string values of varying length up to the maximum specified as the large object length (n[K|M|G]).

The large object length is n, optionally multiple by K|M|G.

Data stored in BLOBs may only be stored in the database and retrieved again, it cannot be used in arithmetical operations.

Specifying the BLOB Length

If you specify <n>K, the length is <n> multiplied by 1 024.

If you specify <n>M, the length is <n> multiplied by 1 048 576.

If you specify <n>G, the length is <n> multiplied by 1 073 741 824.

If you do not specify large object length, the length is implementation-defined as 1 Mb.

Maximum BLOB Length

The maximum length of a BLOB is determined by the amount of disk space available for its storage.

Using BLOBs

You can work with BLOBs as follows:

• Retrieving BLOBs with simple column references in the SELECT clause of a SELECT statement
• Assigning BLOBs using INSERT statements with a VALUES clause
• Assigning BLOBs using UPDATE statements
• Adding BLOB columns using CREATE TABLE or ALTER TABLE
• Dropping BLOB columns using ALTER TABLE
• Altering BLOB column data types using ALTER TABLE

There are some restrictions associated with using BLOBs. The only comparison supported for BLOB values is using the NULL predicate.

A BLOB column may not be part of any primary key, index, unique constraint or primary key constraint.

The comparison restrictions also prevent BLOB columns from being used in DISTINCT, GROUP BY and ORDER BY clauses and UNION statements.

You cannot: use BLOBs in stored procedures, define triggers for tables that contain CLOB columns or add BLOB columns to tables that have triggers.

Numerical

The NUMERICAL data type category contains the following data types:

Data Type	Abbrevi-ation	Description	Range
INTEGER(p)	INT(p)	Integer numerical, precision p.	1 £ p £ 45
SMALLINT	N/A	Integer numerical precision 5.	-32 768 through 32 767 Corresponds to a 2 bytes, signed int.
INTEGER	INT	Integer numerical, precision 10.	-2 147 483 648 through 2 147 483 647 Corresponds to a 4 bytes, signed int.
BIGINT	N/A	Integer numerical, precision 19.	-9 223 372 036 854 775 808 through 9 223 372 036 854 775 807 Corresponds to an 8 bytes, signed int.
DECIMAL(p, s)	DEC(p, s)	Exact numerical, precision p, scale s.	1 £ p £ 45 0 £ s £ p
NUMERIC(p, s)	NUM(p, s)	Exact numerical, precision p, scale s. (Same as DECIMAL).	1 £ p £ 45 0 £ s £ p
FLOAT(p)	N/A	Approximate numerical, mantissa precision p.	1 £ p £ 45 Zero or absolute value 10-999 to 10+999
REAL	N/A	Approximate numerical mantissa precision 7.	Zero or absolute value 10-38 to 10+38 Corresponds to a single precision float.
FLOAT	N/A	Approximate numerical mantissa precision 16.	Zero or absolute value 10-308 to 10+308 Corresponds to a double precision float.
DOUBLE PRECISION	N/A	Approximate numerical mantissa precision 16.	Zero or absolute value 10-308 to 10+308

All NUMERICAL data may be signed.

For all NUMERICAL data, the precision p indicates the maximum number of decimal digits the number may contain, excluding any sign or decimal point.

For decimal data, the scale s indicates the fixed number of digits following the decimal point.

Note: The decimal data with scale zero DECIMAL(p,0) is not the same as integer INTEGER(p).

For FLOAT(p), floating point (approximate numerical) data is stored in exponential form. The precision is specified for the mantissa only. The permissible range of the exponent is -999 to +999.

Note: In Mimer SQL the NUMERIC data type is exactly equivalent to DECIMAL.

Specifying Data Type Precision and Scale

In the following cases, the omission of scale, or the omission of both precision and scale, is allowed (scale may not be specified without precision):

Data Type	Abbreviation
DECIMAL	DEC	is equivalent to DECIMAL(15,0)
DECIMAL(5)		is equivalent to DECIMAL(5,0)
NUMERIC	NUM	is equivalent to NUMERIC(15,0)
NUMERIC(5)		is equivalent to NUMERIC(5,0)

Note: The data type INTEGER is distinct from INTEGER(10).

Datetime

DATETIME is a term used to collectively refer to the data types DATE, TIME(s) and TIMESTAMP(s).

Data type	Description
DATE TIME(s) TIMESTAMP(s)	Composed of a number of integer fields, represents an absolute point in time, depending on sub-type.

DATE

DATE describes a date using the fields YEAR, MONTH and DAY in the format YYYY-MM-DD. It represents an absolute position on the timeline.

TIME(s)

TIME(s) describes a time in an unspecified day, with seconds precision s, using the fields HOUR, MINUTE and SECOND in the format HH:MM:SS[.sF] where F is the fractional part of the SECOND value. It represents an absolute time of day

TIMESTAMP(s)

TIMESTAMP(s) describes both a date and time, with seconds precision s, using the fields YEAR, MONTH, DAY, HOUR, MINUTE and SECOND in the format YYYY-MM-DD HH:MM:SS[.sF] where F is the fractional part of the SECOND value. It represents an absolute position on the timeline.

DATETIME Significance

A DATETIME contains some or all of the fields YEAR, MONTH, DAY, HOUR, MINUTE and SECOND. These fields always occur in the order listed, which is from the most significant to least significant. Year is the most significant.

Each of the fields is an integer value, except that the SECOND field may have an additional integer component to represent the fractional seconds.

For a DATETIME value with a SECOND component, it is possible to specify an optional seconds precision which is the number of significant digits in the fractional part of the SECOND value. This must be a value between 0 and 9. If a SECOND's precision is not specified, the default is 0 for TIME and 6 for TIMESTAMP.

Calendar and Clock

DATE values are represented according to the Gregorian calendar. TIME values are represented according to the 24 hour clock.

Inclusive Value Limits for DATETIME

The inclusive value limits for the DATETIME fields are as follows:

Field	Inclusive value limit
YEAR	0001 to 9999
MONTH	01 to 12
DAY	01 to 31 (upper limit further constrained by MONTH and YEAR)
HOUR	00 to 23
MINUTE	00 to 59
SECOND	00 to 59.999999999

Interval

An INTERVAL is a period of time, such as: 3 years, 90 days or 5 minutes and 45 seconds.

Data Type	Description
INTERVAL	Composed of a number of integer fields, represents a period of time, depending on the type of interval.

There are effectively two kinds of INTERVAL:

• YEAR-MONTH

  containing one or both of the fields YEAR and MONTH.

• DAY-TIME

  containing one or more consecutive fields from the set DAY, HOUR, MINUTE and SECOND.

The distinction is made between the two interval types in order to avoid the ambiguity that would arise if a MONTH value was combined with a field of lower significance, e.g. DAY, given that different months contain differing numbers of days.

For example, the hypothetical interval 2 months and 10 days could vary between 69 and 72 days in length, depending on the months involved. Therefore, to avoid unwanted variations in the downstream arithmetic etc. the variable length MONTH component may only exist at the lowest significance level in an INTERVAL.

The SECOND field may also only exist at the lowest significance level in an INTERVAL, simply because it is the least significant of all the fields.

An INTERVAL data type is a signed numeric quantity (i.e. negative INTERVALs are allowed) comprising a specific set of fields. The list of fields in an INTERVAL is called the interval precision.

The fields in an INTERVAL are exactly the same as those previously described for DATETIME except that the value constraints imposed on the most significant field are determined by the leading precision (p in Interval Qualifiers) for the INTERVAL type and not by the Gregorian calendar and 24 hour clock.

A leading precision value between 1 and the maximum allowed for the field type may be specified for an INTERVAL. If none is specified, the default is 2.

Value Constraints for Fields in an Interval

The table below shows the maximum permitted leading precision values for each field type in an INTERVAL:

Field	Maximum leading precision
YEAR	7
MONTH	7
DAY	7
HOUR	8
MINUTE	10
SECOND	12

The value of a MONTH field, which is not in the leading field position, is constrained between 0 and 11, inclusive, in an INTERVAL (and not between 1 and 12 as in a DATETIME).

Where the SECOND field is involved, seconds precision (s in Interval Qualifiers) can be specified for it in the same way as for DATETIME.

Note that in the INTERVAL consisting only of a SECOND field (INTERVAL SECOND), the SECOND field will have both a leading precision and a seconds precision, specified together.

A seconds precision value between 0 and 9 may be specified for an INTERVAL. If the seconds precision is not specified, a default value of 6 is implied.

Interval Qualifiers

A syntactic element, the interval qualifier, is used to specify the interval precision, leading precision and (where appropriate) the seconds precision.

The interval qualifier follows the keyword INTERVAL when specifying an INTERVAL data type.

The following table lists the valid interval qualifiers for YEAR-MONTH intervals:

Interval Qualifier	Description
YEAR(p)	An interval class describing a number of years, with a leading precision p. It contains a YEAR field in the format: pY.
MONTH(p)	An interval class describing a number of months, with leading precision p. It contains a MONTH field in the format: pM.
YEAR(p) TO MONTH	An interval class describing a number of years and months, with leading precision p. The format is: pY-MM.

The following table lists the valid interval qualifiers for DAY-TIME intervals:

Interval Qualifier	Description
DAY(p)	An interval class describing a number of days, with a leading precision p. It contains a DAY field in the format: pD.
HOUR(p)	An interval class describing a number of hours, with leading precision p. It contains an HOUR field in the format: pH.
MINUTE(p)	An interval class describing a number of minutes, with leading precision p. It contains a MINUTE field in the format: pM.
SECOND(p, s)	An interval class describing a number of seconds, with leading precision p and seconds precision s. It contains a SECOND field in the format: pS[.sF]. (F is the fractional part of the seconds value.)
DAY(p) TO HOUR	An interval class describing a number of days and hours, with leading precision p. The format is: pD HH.
DAY(p) TO MINUTE	An interval class describing a number of days, hours and minutes, with leading precision p. The format is: pD HH:MM.
DAY(p) TO SECOND(s)	An interval class describing a number of days, hours, minutes and seconds, with leading precision p. The format is: pD HH:MM:SS[.sF].
HOUR(p) TO MINUTE	An interval class describing a number of hours and minutes, with leading precision p. The format is: pH:MM.
HOUR(p) TO SECOND(s)	An interval class describing a number of hours, minutes and seconds, with leading precision p and seconds precision s. The format is: pH:MM:SS[.sF].
MINUTE(p) TO SECOND(s)	An interval class describing a number of minutes and seconds, with leading precision p and seconds precision s. The format is: pM:SS[.sF].

Length of an Interval Data Type

The length of an INTERVAL data type is the same as the number of characters required to represent it as a string and is determined by the interval precision, leading precision and the seconds precision (where it applies).

The maximum length of an INTERVAL data type can be computed according to the following rules:

• The length of the most significant field is the leading precision value (p).
• Allow a length of 2 for each field following the most significant field.
• Allow a length of 1 for each separator between fields. Separators occur between YEAR and MONTH, DAY and HOUR, HOUR and MINUTE, and MINUTE and SECOND.
• If seconds precision applies, and is non-zero, allow a length equal to the seconds precision value, plus 1 for the decimal point preceding the fractional part of the seconds value.

ROW Data Type

There is an additional data type supported by Mimer SQL, called the ROW data type used in stored procedures.

A variable which is declared as having the ROW data type implicitly defines a row value, which is a single construct that has a value which effectively represents a table row.

A row value is composed of a number of named values, each of which has its own data type and represents a column value in the overall row value.

A ROW data type can be defined either by explicitly specifying a number of field-name/data-type pairs or by specifying a number of table columns from which the unqualified names and data types are inherited.

A ROW data type definition can be specified where one of the above data types would normally be used in a variable declaration in a compound statement, see the Mimer SQL Programmer's Manual, chapter 12, The ROW Data Type, for details.

ROW Data Type Syntax

The syntax for defining a ROW data type is:

The following points apply to the specification of a ROW data type:

• A domain cannot be specified for data-type.
• The value specified for data-type can be a ROW data type specification.
• Two fields in the same ROW data type specification must not have the same name (this restriction applies equally to fields named by specifying a field-name value and those named by inheriting the unqualified name of a table column).
• If table-name is specified without a list of column names, all the columns in the table are used to define fields in the ROW data type.

The NULL Value

Columns which contain an undefined value are assigned a NULL value.

Depending on the context, this is represented in SQL statements either by the keyword NULL or by a host variable associated with an indicator variable whose value is minus one, see the Mimer SQL Programmer's Manual, chapter 7, Indicator Variables.

The NULL value is generally never equal to any value, not even to itself. All comparisons involving NULL evaluate to unknown, see Comparisons.

Note: NULL values are treated as equal to each other for the purposes of DISTINCT, GROUP BY, and ORDER BY.

NULL is sorted at the end of ascending sequences and at the beginning of descending sequences.

Data Type Compatibility

Assignment and comparison operations generally require that the data types of the items involved (literals, variables or column values) are compatible but not necessarily exactly equivalent.

Any exceptions to this rule are specified in the detailed syntax descriptions in SQL Statement Descriptions.

All character data is compatible with all other character data.

Numerical data is compatible with other numerical data regardless of specific data type (integer, decimal or float). Rules for operations involving mixed numerical data types are described in Comparisons.

Datetime and interval data types can be combined in arithmetic operations, for details, see Datetime and Interval Arithmetic.

Values stored in host variables (but not literals or column values) may be converted between character and numerical data types if required by the operation using the variable. The declared type of the variable itself is not altered.

Similarly, character columns may be assigned to numerical variables and vice versa. The rules for data type conversion are given below.

Variables may be converted between different data types by using the CAST function.

Datetime and Interval Arithmetic

The following table lists the arithmetic operations that are permitted involving DATE, TIME, TIMESTAMP (DATETIME) or INTERVAL values:

Operand 1	Operator	Operand 2	Result Type
DATETIME	-	DATETIME	*See discussion below.
DATETIME	+ or -	INTERVAL	DATETIME
INTERVAL	+	DATETIME	DATETIME
INTERVAL	+ or -	INTERVAL	INTERVAL
INTERVAL	* or /	NUMERIC	INTERVAL
NUMERIC	*	INTERVAL	INTERVAL

Operands cannot be combined arithmetically unless their data types are comparable, see Comparisons. If either operand is the NULL value, then the result will always be the NULL value.

If an arithmetic operation involves two DATETIME or INTERVAL values with a defined scale, the scale of the result will be the larger of the scales of the two operands.

When an INTERVAL value is multiplied by a NUMERIC value, the scale of the result is equal to that of the INTERVAL and the precision of the result is the leading precision of the INTERVAL increased by 1. In the case of division, the same is true except that the precision of the result is equal to the leading precision of the INTERVAL (i.e. it is not increased by 1).

When two INTERVAL values are added or subtracted, the scale (s) and precision (p) of the result are described by the following rule:

 p = min(MLP, max(p'-s', p"-s") + max(s', s") + 1)
 s = max(s', s")
 

where MLP is the maximum permitted leading precision for the INTERVAL type of the result, refer to the table in Interval for these values.

The interval precision of the result is the combined interval precision of the two operands, e.g.

 DAY TO HOUR + MINUTE TO SECOND
 

will produce a DAY TO SECOND result.

One DATETIME value may be subtracted from another to produce an INTERVAL that is the signed difference between the stated dates or times.

The application must, however, specify an INTERVAL date type for the result by using an interval-qualifier.

Thus, the syntax is:

 (DATETIME1 - DATETIME2) interval-qualifier
 

Example:

 (DATE '1996-01-09' - DATE '1996-01-01') DAY
 

This, therefore, evaluates to INTERVAL '8' DAY.

Data Types for Parameter Markers

Parameter markers in statements submitted to dynamic SQL are assigned data types appropriate to their usage. See the Mimer SQL Programmer's Manual, chapter 11, Dynamic SQL, for a discussion of dynamic SQL.

For parameter markers used to represent NUMERICAL data in arithmetic or comparison expressions, precision 45 is used.

For parameter markers used to represent data assigned to columns, the precision is assigned in accordance with the column definition.

Host Variable Data Type Conversion

When a host variable is used in assignments, comparisons or expressions where the data type of the variable is different from the data type of literals or column declarations, an attempt is made internally to convert the value of the variable to the appropriate type.

Conversion between a character variable and a character value, is always allowed. The conversion follows these rules:

• When assigning a character value to a character variable, where the variable is longer than the character value, the variable is padded with trailing blanks.
• When assigning a character value to a character variable, where the value is longer than the variable, the value is truncated and a warning status is returned. If only blanks are truncated, no warning is returned.
• When assigning a variable length character, i.e. a VARCHAR or NCHAR VARYING, column from a character variable, the column is padded with blanks up to the length of the character variable if the column is longer than the variable.
• When assigning a variable length character column from a character variable, where the column is shorter than the variable (except for trailing spaces), the assignment will fail and an error message is returned.
• When assigning a national charact