Will the current version of SYNCSORT provide the functionality to handle this problem?
I'm not sure. . . The free format of the data adds complexity. If the data in the "built" records was fixed, it would be possible (how to do that is already posted<g>). Also, in the receiving program, if the data is variable, how will the "receiving" cobol code know where to break the individual fields? There aren't any delimiters.
If there are 240 files and many of these may have 300 records, i believe the overall record length will soon need to be dealt with also. You might want to run a process to identify the largest possible collection for an ssn.
How is the single file of all of the component pieces going to be build as input to this "sort"? If there is no single input file planned, how will all of the 240 sets of data get into this process (unless they all have the same dcb info, concatenation probably won't work)?
You too have a great weekend - i'll check in to see what's up,