We were able to work around this restriction by running an ETL job through which a superuser copies the contents of updated system table records into a publicly accessible scheme. Far from perfect, but I’ve found it to be a manageable solution.

That’s my experience as well.

1 to this request

Just ran into this issue again. I don’t see why a feature which is being applied to a SQL results set should be limited in any way whatsoever by the method in which said results set is generated.

I would say that not only is this an important feature to make available to end users, but also that given how frequently we need to make use of table functions in order to accommodate pivot-wise calculations that are not supported in the model, limi...