DSXchange

You can't use 'transaction grouping' if you want an 'all or nothing' style load. It forces the transaction size to 1 so that each 'unit of work' is either commited or rolled back together.

If you want all or nothing, uncheck the grouping option and use a transaction size of 0 for all links.

Pivot stage.

Substring ( [] ) or OConv pairs with the proper coversion codes & masks, then concatenate with a dot between would do the trick. Assuming an ISO standard timestamp format: Convert("-","",YourTimestamp)[5,4] : "." : Convert("-","",Your...

Been posted before - try searching for "check that 'dbx' and 'sed' are installed and on your PATH" and see what turns up.

I searched for 'registry' to find that one, but 'menu bar' found 20 matches (including the link I posted) when I did an Exact Search. Ah well, at least the silly thing is back.

Should be the same name you used as the 'Database source name' in the OCI stage, a valid entry in your tnsnames.ora file.

As noted above, set the data element to 'XML' for that field and it will pass it through unchanged.

Because the first action has to fail to trigger the second. Without a unique index over your keys, Case I never fails and all records insert. Case II only inserts when the update effects zero records.

Google?

There would be no 'performance issue' with doing this in a script or after-job routine.

To do it in the job it must be a job parameter, then you can use it directly in the filename.

You need to check the link count after the fact in a separate process. This would typically be done in a Sequence job - a Routine Activity stage to do the DSGetLinkInfo and then triggers to send a zero result to a Notification Activity stage.

Me three! :D

Or the 8.x release, I do believe.

Post your syntax.

Check out in the Manager or search here for discussions on the 'KeyMgtGetNextValue' routine, that's a good starting point.