On Fri, 23 Jun 2017 09:57:10 +1000, Andrew Rowley wrote: >The biggest problem I think is proving the conversion is "round trip". > >I would suggest doing all conversion on the mainframe, and if you need >to be 100% sure you could add a step to do the conversion back into a >temporary file and compare it with the original. > Yes. And SuperC may be a better tool than checksum because if the checksum treates the data as a stream it may be oblivious to changes in record boundaries.
>Any conversion on another platform has a significant risk that you can't >convert the result back to the mainframe format. The risk still exists >with the mainframe conversion, but at least you can add the comparison >step to catch it. > But the "round trip" test should include transfer to the "distributed platform" and retrieval from it to demonstrate that those steps are valid. And the checksums generated on the mainframe both before and after encryption should be compared with those on the "distributed platform" after and before decryption. Such checksums should routinely be preserved for possible use in troubleshooting. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
