Originally posted by: SystemAdmin

We are implementing maps that need to produce fixed length records in flat files (on the mainframe). When such a file is allocated with a wrong record length by the LRECL-value, the resulting file is messed up as the contents of records get shifted to the left or right.
When a COBOL program is confronted with a wrong record length we were used to an abend about a mismatch and this is exactly what our supporting organisation expects from Datastage TX as well, so they are alerted in time in case of a JCL-error.
How can we cope with this support-requirement ?
#DataExchange
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender

Originally posted by: SystemAdmin

Hi Peter,

I'm not sure if I understand the requirement correctly, but perhaps the following will help.

• Pass the file through a type tree that it should fit through, including fixed record sizes and even component rules to check the data.

• Or use the SIZE command in the output card after the card that produces the record, to check that the size of the record/file is what you expect it to be. If it is incorrect, FAIL the map with a custom error message.

In either case, the error will be highlighted in the map audit log, which you can process to alert the Support team. Do you already have an error handling mechanism in place for TX?

Regards,

Tarun
#IBMSterlingTransformationExtender
#DataExchange
#IBM-Websphere-Transformation-Extender

Originally posted by: SystemAdmin

Are you padding your output fields to fixed length?
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender
#DataExchange

Originally posted by: SystemAdmin

I'm afraid I didn't describe my issue clearly. I'll clarify:

1. We are capable of developing a map that produces an output file with fixed length records and we are capable of allocating the resulting mainframe file with the correct record length (through LRECL), but our support group wants to be alerted when a mismatch occurs between the record length in the map and the allocation of the output file (as a result of a human error).

2. We intend to do rudimentary error handling: when a type error occurs we make a calculation based on the "Total object count" and the "number of objects per line" to determine the error location in the input file. So, we don't use the restart attribute as we consider each input error fatal.
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender
#DataExchange

Originally posted by: SystemAdmin

You might want to read in the file created into another map, using the same tree that created the file, and if it fails validation, send an e-mail or other notice to your support group.
#DataExchange
#IBMSterlingTransformationExtender
#IBM-Websphere-Transformation-Extender

Originally posted by: SystemAdmin

Reading the created file by another map wouldn't help, because Datastage is not aware of any record-seperation in a fixed record length file, so the created file will be valid independent of the actual record length.
#DataExchange
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender