On Mar 15, 2010, at 9:38 , Felix Meschberger wrote: > Hi, > > On 15.03.2010 09:25, Marcel Offermans wrote: >> On Mar 15, 2010, at 9:09 , Felix Meschberger wrote: >> >>> I have an incertainty about Service PID syntax as well as enforcement of >>> it in the ConfigurationAdmin service. >>> >>> As specified in Section 5.2.6, Persistent Identifier (PID), a PID is >>> defined the same as a symbolic-name as defined in Section 1.3.2, General >>> Syntax Definitions (all in the R 4.2 Core Spec). Such PIDs are referred >>> to in the Configuration Admin specification to identify configurations. >>> >>> My questions: >>> >>> (1) One of the examples in Section 104.3.1.3, Devices, of the >>> Compendium spec seems to violate this restriction proposing >>> the string "802-00:60:97:00:9A:56" as a PID (this is invalid >>> because colons are not allowed in symbolic names). >>> Is this a bug or intention ? >> >> Both 5.2.6 and 104.3.1 mention that a PID *SHOULD* (not MUST) follow the >> symbolic-name syntax. > > Oh ! I missed that point ... > > Thanks. > > Thus, if Secion 104.4.2, Configuration Properties, in the compendium > spec says "The format of a property name should be:" it actually means, > may or may not be ... > > Hmm, then the Felix Configuration Admin implementation would actually be > too strict with respect to validating property keys ... (we throw if the > actual property format does not match the definition)
That is how I would interpret the specs, yes, but I would not mind hearing the opinion of others on this. Greetings, Marcel _______________________________________________ OSGi Developer Mail List [email protected] https://mail.osgi.org/mailman/listinfo/osgi-dev
