Thanks Mark.
Having carried out some brief testing it looks like pipelines don't work for the sandboxed application.
Each time a sandbox is created a new set of DBIDs are generated, so testing in the sandbox with pipelines active would seem to involve:
1. Creating a token for the sandbox (as the token for the live application doesn't provide access to the sandbox).
2. For every table referenced in pipelines, retrieving the new DBIDs from the sandbox.
3. Mapping these to the DBIDs in the live application.
4. Exporting the YAML for the pipelines.
5. Replacing DBIDs.
6. Replacing references to the token.
7. Modifying the name of the pipeline to distinguish it from the original.
8. Importing the YAML.
I'm not sure if it's possible to derive the 7 character alphanumeric reference to the token that appears in the YAML without - as I did - creating another pipeline and exported the YAML from that, so there's possibly another step to add.
------------------------------
Jeremy Anson
------------------------------