Tom wrote:
> I do find it interesting that we now have two reports of somebody
> doing "ALTER ROLE SET role = something". In the older thread,
> I was skeptical that that had any real use-case, so I wonder if
> Charlie has a rationale for having done that.
Unfortunately I haven't heard back from the original developer
who set up this role configuration, but if I do then I will share
their intentions. In any case the invalid configuration had been
removed from every other role except one (certainly by mistake)
which lead to me rediscovering this issue.
I tested the above patch with the invalid data locally and it avoids
the restore error that we ran into previously. Also it requires no
intervention to progress with pg_upgrade unlike my initial idea of
adding an check, so it is definitely simpler from a user perspective.
Thank you for taking a deep look into this and finding a better
solution.
Best regards,
Charlie Hornsby