New features for new OSes (part 3)

Advanced profiling for user and system configuration

Configuration should be classified in three fields: system configuration, users configuration and application configuration. These three sets can eventually intersect and share some data (for example, an application could check system configuration for printing or current user configuration for colour preferences), so the configuration system should be able to permit partial access to some data to some applications. This means that the configuration system should be much more intelligent than – for instance – a registry file, and some database capabilities will definitely come up in this context.

In a network environment, an advance configuration system should also provide the capacity to centralize system administration (in collaboration with the auto-update and auto-heal system) and to remote user configuration and permissions, so that the machines attached to the network are able to download on demand whatever the logged user needs.

How would be possible to obtain this behaviours? With a DBMS managed configuration system, we could identify three entities (systems, users and applications): each entity has its own credentials (in commercial systems this could also include licence checking) and it is stored in the configuration database. The database engine is able to retrieve informations also from the centralized server to obtain network administration and user distribution abilities; there are very strict rules on how entities can access data: for example an application executed from a certain user can only gain access to this specific user configuration, and is permitted to write configuration informations concerning only the “this user, this application” join; the very same limitations are applied in the “this application, this system” join.