The SORTIN file may consist of concatenated data sets, up to the limit supported by the operating system. (Note: If you want to combine VSAM and non-VSAM files as input to a SORT or COPY, use the MULTIIN facility described in section Multiple Input Files.)
MFX must determine one set of DCB characteristics to use for reading all data sets in the concatenation. The following rules apply to the DCB characteristics:
-
When the first data set is fixed-length (RECFM=F, FB, FBS), all subsequent data sets must be fixed-length and have the same LRECL.
-
When the first data set is variable-length (RECFM=V, VB, VS, VBS), all subsequent data sets must be variable-length.
-
For variable-length data sets, the LRECL of the first data set is used except for the following situations:
-
The LRECL of a subsequent data set is used if that LRECL is the largest found and is available at sort initialization. An LRECL is available at initialization if it is specified on a SORTIN DD statement or exists in the label of a SORTIN disk data set.
-
A record length specified via the l1 value on the RECORD control statement is used if it is the largest record length found.
-
-
For both fixed and variable-length data sets, the BLKSIZE of the first data set is used unless the BLKSIZE of a subsequent data set is the largest found and is available at sort initialization. A BLKSIZE is available at initialization if it is specified on a SORTIN DD statement or exists in the label of a SORTIN disk data set.
The following shows sample JCL for concatenating input data sets:
In the preceding example, one disk and two tape data sets have been concatenated. Any one of these data sets could be presented first. Position is not dependent upon BLKSIZE or LRECL. If the LRECL or BLKSIZE cannot be determined at SORT initialization, the first data set must carry the largest LRECL or BLKSIZE of the concatenation. Typically the LRECL or BLKSIZE cannot be determined when the input consists of concatenated tape data sets and the JCL lacks a DCB specification.