Current, pum works by providing service names, which are defined in pg_service.conf.
This is not too practical in some scenarios :
from with QGIS, where you may have your database service/credentials defined in the browser already
from the command line, in case you're just running a one off command
docker setup, where you'd need to mount the pg_service file somehow
not necessarily very secure, as you need to remember the file afterwards if you don't want to store the unencrypted admin password
I think it should be relatively easy to adapt pum to be able to use connection strings as defined in https://www.postgresql.org/docs/10/libpq-connect.html#id-1.7.3.8.3.2
This would still allow to use pg_service conf when useful, and regular connection parameters otherwise.
Current, pum works by providing service names, which are defined in pg_service.conf.
This is not too practical in some scenarios :
I think it should be relatively easy to adapt pum to be able to use connection strings as defined in https://www.postgresql.org/docs/10/libpq-connect.html#id-1.7.3.8.3.2 This would still allow to use pg_service conf when useful, and regular connection parameters otherwise.
Or was there a reason to primarily rely on this ?