On my team, we've been implementing error handling in agenda jobs in one of two ways:

1. Catch the error, call job.fail(err); job.save(), return
2. Throw the error

After looking at the job processing code for the run method, I confirmed the two approaches were equivalent.

While

The documentation website linked in this repository description seems to have moved. When clicking the link, it shows the question "how do you get here?".

Right now to import CSV file one should use:

• psql /copy SHELL task for client-side files
• COPY FROM SQL task for server-side files

Server-side files are rare so it can be ignored.

Test for code examples in the documentation are run in continuous integration through Sphinx, but it would be preferable to have them run when executing pytest.