>>-ISPEXEC--TBSORT--table-name--FIELDS(sort-list)--------------><

>>-CALL--ISPEXEC--(buf-len,--buffer);--------------------------><

or

>>-CALL--ISPLINK--('TBSORTbb'--,--table-name,--sort-list);-----><

table-name

Specifies the name of the table to be sorted.

sort-list

Specifies sort fields. The syntax is as follows:

(field-name1, B|C|N|Yn, A|D,field-name2, B|C|N|Yn, A|D, …)

Each sort field-name must be either a KEY or NAME field. The first (left most) field-name is the primary key (most significant) and the rows are then collated based on the values of the field-names.

The field-name is followed by a sort field type designator. The sort field type designator can have a value of 'C' for a character sort, a value of 'N' for a numeric sort, a value of 'B' for a binary sort, or a value of 'Yn' for a year sort. For English, where sorting is in EBCDIC sequence, specifying either C or B as the sort field type designator causes the same sort order. For other languages, where the character format can be other than EBCDIC, only B is to be specified for a binary sort.

The 'Yn' sort is treated as a character sort where the variable being sorted is a date variable, with n being the offset of the beginning of a 2-byte year in the variable. To sort a table into a valid ascending or descending date sequence, the date field must have a format with the most significant part (the year) at the start, the least significant part at the end, and a sort field type designator of Y1. Some examples of valid formats are:

YY/MM/DD
YYDDD
YY-MM

Internally, TBSORT expands the year to a 4-byte year using the ISPF configuration table field YEAR_2000_SLIDING_RULE to calculate the century value.

The collating sequence for character sorts during DBCS and English sessions is in EBCDIC order. This means, for example, that all lowercase letters precede uppercase letters when sorting in an ascending sequence. For other languages, a character sort is done such that both uppercase and lowercase, as well as accented and non-accented versions of a letter, are sorted in the proper order.

The sort field type designator is followed by a sort sequence direction value. The sort sequence direction value can be either 'A' (ascending) or 'D' (descending). The field type designator and the sort sequence direction can be omitted for the last named field only. They default to 'C' (character) and 'A' (ascending), respectively.

In some languages, the comma is used in place of a decimal point. To accommodate different usages, three numeric representations are supported: period, comma, and French representations.

Table 1. Decimal Point Representations

Convention	Example	Example	Where Used
Period	1,234.56	0.789	Japan, Mexico, UK, USA
Comma	1.234,56	0,789	Most other countries
French	1234,56	0,789	France, South Africa

The TBSORT service accommodates these three numeric representations. The convention used is determined by the language of the session, specified by the value of ZLANG in the system profile table. The current English version accepts only the period, treating it as the delimiter of the whole and decimal portion of a number. Sorting is based on the specified language convention.

These restrictions apply to fields for a "numeric" type sort:

1. The field must be a decimal number and optionally can contain a plus (+) or minus (-) sign. The decimal number can be either a whole number (for example, 234) or a mixed number (for example, 234.56), composed of a whole number followed by a decimal point. A decimal point is not required after a whole number, but is required in a mixed number. (Under the period convention, the decimal point is represented by a period (.); under the comma or French conventions, the decimal point is represented by a comma (,).) No other characters are allowed except leading blanks.
2. No numeric string can exceed 16 characters. This length value includes any plus or minus sign, any blanks, or a decimal point.
3. The largest value that can be sorted is plus or minus 2 147 483 647.
4. The string can have leading blanks following the sign character.

buf-len

Specifies a fullword fixed binary integer containing the length of "buffer".

buffer

Specifies a buffer containing the name of the service and its parameters in the same form as they would appear in an ISPEXEC call for a command procedure.