My suggestion is to write two routines. One is to export the data to two flat files (one to contain data the other to contain a list of files with associated item and byte counts) the other routine to read one of the flat files to determine the files to create and their correct sizes and or type the process the other flat file by writing the data, item by item into the appropriate file(s).
This we have done many times and works regardless of the source and destination system you are involved with. It also has the added advantages of allowing you to automatically check the import has worked OK, that you can filter out unwanted characters or pieces of data such as object code from dictionaries or CHAR(0)s as is often found in some old systems PLUS when it comes time to do the final cutover you can modify the original SELECT list so you only need to bring across that which has changed and is required. All you set up and modifications needed to get the software to work on the destination machine need not be done again as you are simply loading the data into files.
Open Systems Professionals
----- Original Message -----
-- u2-users mailing list [EMAIL PROTECTED] http://www.oliver.com/mailman/listinfo/u2-users