The permissions =============== Permission's names ------------------ The proposal here is to specify a naming scheme for permissions that allows the system to be as stateless as possible. The current specification includes in the naming of permissions either the name of the bound binding when existing and the level of the permission itself. Doing this, there is no real need for the framework to keep installed permissions in a database. The permission names are [URN][URN] of the form: urn:AGL:permission::: where "AGL" is the NID (the namespace identifier) dedicated to AGL (note: a RFC should be produced to standardize this name space). The permission names are made of NSS (the namespace specific string) starting with "permission:" and followed by colon separated fields. The 2 first fields are `` and `` and the remaining fields are grouped to form the ``. ::= [ ] ::= 1* ::= | | | ::= "-" | "." | "_" | "@" The field `` can be made of any valid character for NSS except the characters colon and star (:*). This field designates the api providing the permission. This scheme is used to deduce binding requirements from permission requirements. The field `` can be the empty string when the permission is defined by the AGL system itself. The field `` if starting with the character "@" represents a transversal/cross permission not bound to any binding. ::= 1* The field `` is made only of letters in lower case. The field `` can only take some predefined values: - system - platform - partner - tiers - owner - public The field `` is made of `` separated by colons. ::= 0*(":" ) The names at left are hierarchically grouping the names at right. This hierarchical behaviour is intended to be used to request permissions using hierarchical grouping. [URN]: https://tools.ietf.org/rfc/rfc2141.txt "RFC 2141: URN Syntax"