-
Notifications
You must be signed in to change notification settings - Fork 1.6k
Description
Is your feature request related to a problem? Please describe.
When an error occurs, it's useful to have a QueryJob object to inspect the errors properties, as well as check for other details about the job that may have caused the failure. At the very least, a job ID is needed if one wants to file a support ticket about the failure.
Describe the solution you'd like
As identified in #9067 (review), when an error occurs with a %%bigquery magic, it would be useful to save the query job object for further inspection of the error and the rest of the job.
A problem with this is that in successful cases destination_var will be a pandas DataFrame, but in failure cases destination_var will be a QueryJob object. Since the magics are primarily used interactively, this probably isn't that much of a problem.
Describe alternatives you've considered
Print out the job ID and have the user call client.get_job() in order to inspect the object. This does not work in the case of dry run queries, as no job resource is saved in the backend.