Re: [salud] New version of the ABNF-syntax

[[e-mail] (Dale R. Worley)]

> From: Paul Kyzivat <[e-mail]>
> 
> Yes. The thing is that this does not define any normative behavior.
> 
> The person that inserts the URN can put in anything they like, and then 
> it means what it means. If you inserted a URN without a date, then it 
> has some meaning, even if it is not the meaning that you intended.
> 
> E.g. when you insert a URN you may omit the date, thinking that the 
> current owner is the first owner. Its meaning is then defined by the 
> first owner. If the current owner is not the first owner, then the only 
> error is that you referred to a different meaning than you intended.
> 
> And if the person that defines the actual alert that this URN maps to 
> makes the same mistake, then all is well.

That is, I think, an unavoidable problem here.  We can set up a
framework that defines who is permitted to "define the meaning" of a
URN, but "defining the meaning" of a URN has no conseqences at the
level of the operation of the protocol.  All we are doing is providing
guidance to the implementers who they should ask to learn the social
content of a particular URN.  (And we define a bunch of rules
regarding how certain URNs relate to other URNs, which URN is the most
important when processing a set of URNs, etc.)

This is somewhat like priority levels.  We define "emergency",
"urgent", etc., but there are no protocol consequences of those words,
their meaning has an effect only at higher levels of processing.

Dale