Martin, I think you you consider the REFORMAT record, things will be clearer. In the JNFnCNTL files the source records are extended to 501 bytes, with the A/B value being the low-order part of the key (to ensure that there will be no match.
But, in the REFORMAT statement, only the 500 bytes of original data are included (plus the match-marker). The match-marker happens to be in the position where the "A" value would be, if it had been included in the REFORMAT record. z/VSE and its forerunners don't concatenate, as Frank explained. So there you name a "DD" (a DLBL) for each input dataset. SORTIN01, SORTIN02, etc. Yes, on z/OS DFSORT the SORTINnn are used for MERGE. On z/VSE they are not. As far as I can tell, the z/VSE DFSORT is... somewhat behind the z/OS DFSORT. Frank, if you've not been using z/OS DFSORT for long, it is radically different from z/VSE and well worth investing time in. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
