DSXchange

In that case a sequential file stage will be easier to use.

The parameter values used for a run are visible at the head of the log entries. Does the second run use the correct values?

The files generated for BASIC sources during the compilation are put into the RT_BPnnn.O directory. Since they are not object coded files they don't need to be loaded or otherwise changed in order to run. The "pseudocode" mentioned are these files, and the assembler-like mnemonics in human-readable ...

You can also do this with the normal sequential file stage. Could you explain - does the file have both ASCII and EBCDIC columns?

No, you do not need to put in a dummy stage.

Somesh - this post belongs in the Parallel forum as it is specific to that product. The message states that the PEEK operator has a column called "ROW_ID" which does not exist in the input data.

The "pseudocode" I am referring to is created in the {name}.O directoryand, as Ray D. has already pointed out, the contents can be viewed using the VLIST command.

Try resetting the job in the director and checking the entry "From previous run..." in the log.

Are you certain you are reading this as pipe-delimited 6 fields? If yes, then the remove duplicates using the 2nd field will work.

RX? Nothing is proof against a true professional DataStage idiot.

No, that cannot be done using the command line export utility.

are you certain that you have an XML file (edit it and check the format)? If yes, then you have a corrupted format; if no, then you have named a .dsx file incorrectly. Could you post the first couple of lines of your export file if you have further questions?

For me "DS Batch" and "Job Sequence" are synonymous. You can use the user status to transfer a value from the called job to the controlling Job Sequence and then pass that to another job.

Server is easier to spell, too.