The smallest unit of data is a value. A value can result from several different origins: a column, a constant, a function, an expression, a host variable, or a sub-query. CONNX interprets data based on its data type.

CONNX supports the following SQL data types:

CONNX maps these SQL types to the most appropriate Native type for each data adapter.

All data types can be set to SQL NULL, which is not the same as a string of zero length or a numeric 0.

SQL Data types VARCHAR, LONGVARCHAR, VARNCHAR, and LONGVARNCHAR are variable-length character-strings. These data types contain character sequences; the character-string length is determined by either the data type definition or from the value itself. The maximum length of the data type is derived from the definition of the data type; if the value originates from a column which has been defined as a variable-length character-strings with length 15, the maximum length is always 15, but the length of a particular value may be between zero and 15.

SQL Data types CHAR and NCHAR are fixed-length character strings. The length of a value with the data type fixed length character-string is determined by the definition of the origin of the value; if the value originates from a column which has been defined as a fixed length character-string with length 15, a value originating from this column will always have a length of 15.

CONNX supports two categories of character data types: CHAR (single byte), and NCHAR (multibyte/UNICODE).

CHAR: Single byte character types use a single "byte" (a value ranging from 0 to 255) to represent each character in the string. With Latin based alphabets for example, the CHAR data type is sufficiently large enough to represent all the characters in the alphabet.

NCHAR: For non-Latin languages such as Chinese and Japanese, there are more than 256 characters in the alphabet. The NCHAR data type, where every character is stored in two bytes (a value ranging from 0 to 65535) is used to represent all the characters in these alphabets. This is also referred to as utf-16, or UNICODE.

Numeric data types are used to specify the representation form of numeric values. CONNX supports four different representation forms of numeric values. Each numeric value's form of representation has a precision and a sign.

In addition, the decimal forms of representation have a scale. The scale of a numeric value is defined as the number of digits in the fractional part of the number. The scale cannot be larger than the precision nor can it be negative.

The maximum precision/scale of a particular datasource may be much smaller than the CONNX maximum.

CONNX supports BINARY, VARBINARY and LONGVARBINARY binary data types. The data stored within the binary data type is not byte swapped nor interpreted in any way.

For Adabas, the binary data type behavior deviates from the ANSI standard. The Adabas binary data type is an unsigned integer and is subject to byte swapping where appropriate. Application programs can use multiple interpretations when accessing the Adabas binary data type; in most cases the Adabas binary data type will be interpreted as a bit pattern. Adabas Binary values have a maximum length of 126 bytes. The maximum possible number of bits is 1008.

CONNX is capable of converting a value of a certain data type to another data type. Precision or length may be lost if the data type is converted into a type with smaller precision/length.