The use case for having the openid-server as an external client application was so that multiple openid-servers could be run against a single Crowd server.

This fringe use-case benefit pales in comparison to the other benefits of having the openid-server built-in to the crowd-console:

• easier configuration (administration can be done via a single console)
• easier backups (1 database, not 2)
• cleaner code (no need to push things to crowd-core to make them available to openid-server)

This will require role permissioning within the Crowd application (ie. crowd-users, crowd-administrators, possibly crowd-openid-users, etc).