FastSort Manual

Using FastSort System Procedures
FastSort Manual—429834-003
5-23
SORTMERGESTART Procedure
in-file-count input
INT(32):ref:*
is an array including one 32-bit entry for each input file. Each entry contains the
maximum number of records in the corresponding input file. When input is from a
source other than disk, SORTPROG uses in-file-count to estimate the space
required for the scratch file.
If you omit in-file-count or specify –1, SORTPROG determines the maximum
number of records as follows:

For a structured disk file, SORTPROG estimates the number of records in the
file by looking at the end-of-file location and determining the structured
overhead.

For an unstructured disk file, SORTPROG calculates an approximate number
of records in the file. The approximate number of records for an EDIT file is the
end-of-file location multiplied by 2 and divided by the record length. The
approximate number of records for other unstructured files is the end-of-file
location divided by the record length. The default record length for unstructured
files is 132 bytes.

For files other than disk files and records supplied by SORTMERGESEND, the
default number of records is 50,000.
in-file-record-length input
INT:ref:*
is an array including one 16-bit entry for each input file. Each entry contains the
maximum record length in the corresponding input file. The largest record length
allowed is 4080 bytes. If you omit in-file-length or specify –1, SORTPROG
uses the default record length.
You can omit this parameter when the input file is a structured disk file, because
the length is in the file label.
To use an odd unstructured file for an input file, you must specify the correct length
in in-file-record-length. For unstructured disk files, files other than disk
files, and records supplied by SORTMERGESEND, the default maximum record
length is 132 bytes.
format input
INT:value
specifies the output record format with one of these values: