IBM Mainframe Forum

Trying to write a rexx program in which i compare a portion of a record from fileA (flat file) against a portion of a record from all the records in fileB (flat file). If no match is found i then read the next record from fileA and compare it to all the records in fileB. The problem is once all the records in fileB have been read once I can't seem to read the file again. I know the rc goes from 0 to 2 once all the records are read and this causes the EOF marker to go from 'no' to 'yes'. However i do not know how to reset this so fileB can be read multiple times (until all records from fileA have been read).

pretty primitive approach

when a match is found what happens ???
are You going to modify file A or file B ???

if You describe the requirement rather than your attempt to a solution You might receive better suggestions

No need to get insulting. This is my first time posting a question. When a match is found the record from fileB that has the match is copied over to a separate file.

How big are your files? Are they sorted? What have you tried in the area of closing and re-opening the file? Have you considered using JOINKEYS within your sort product?

FileA is about 1200 records. FileB is approx 35,000 records. They are not sorted. I am taking an eight byte field from each record in fileA and comparing it to an eight byte field in each record in fileB. When a match is found in fileB I move a copy of that record over to a separate outfile. The record in fileA must be compared to 'every' record in fileB. Then the next record in fileA is compared to every record in fileB - and so on. I guess my problem is how do I get the fileB dataset reopened after a record from fileA has has gone through all the fileB records. The fileA records are only read once. thx

They are not sorted.

The first step of your rexx program should be to sort the files.

Based on the sizes of the files, I would try to only read FileB only one time.

You should index FileA something like this, which builds an array of keys in fileA:

Then as you read in records from FileB, see if there is a match:

It sounds like you wanted to read fileB 1200 times:
1200 x 35000 = 42 million records

I think that would take you many minutes of elapsed time.

Both fileA and fileB could be broken down into smaller chunks if that would make any difference. I'm getting the feeling that in rexx a sequential file can only be read through once which is interesting because I know LMM services will allow you to read through all the members in a pds multiple times.

However i do not know how to reset this so fileB can be read multiple times (until all records from fileA have been read).

This is a technical forum. The details of the program are important. Show us the code that reads the file until EOF and then tries again. And show us a trace.

I suspect you need to read about the FINIS parameter of EXECIO. Or possibly, use the linenum parameter (after the DD name):

With those small numbers of records you could read them all into 2 stems and then process stem a against stem b