Re: [core] Ordering question for draft-hartke-t2trg-coral

[Klaus Hartke <[e-mail]>]

Hi Jim,

> In the process of working on this, I am trying to determine if there is a
> real ordering to the items in a document or not.  The only reason that I can
> see for having one is the possibility directives change the current context.
> If one stores all of this as absolute paths (or relative to a known point),
> one could synthesis the base directive on the fly when doing serialization
> and thus the order of links and forms would not be significant.

in the serializations, elements clearly have an order on the wire, and
the meaning of each elements depends on the current processor state
("current context") that changes based on the directives processed so
far:

    #base <coap://foo.example/>
    quux </1>
    #base <coap://bar.example/>
    quux </2>

is not equivalent to

    #base <coap://foo.example/>
    quux </2>
    #base <coap://bar.example/>
    quux </1>

In the data model, the question is valid whether a link `quux
<coap://foo.example/1>` followed by a link `quux
<coap://bar.example/2>` is semantically equivalent to the link `quux
<coap://bar.example/2>` followed by the link `quux
<coap://foo.example/1>`:

Is

    #base <coap://foo.example/>
    quux </1>
    #base <coap://bar.example/>
    quux </2>

equivalent to

    #base <coap://bar.example/>
    quux </2>
    #base <coap://foo.example/>
    quux </1>

?

This question is currently left open in draft-hartke-t2trg-coral-09.
I'd say the answer mostly depends on the semantics of the link
relation types. E.g.,:

* a link to the mother of a person followed by a link to their father
is probably equivalent to a link to their father followed by a link to
their mother;

* a link to a resource with sz 100 followed by a link to a resource
with sz 50 is probably not equivalent to a link with sz 50 followed a
link with size 100 if the client requests the links be sorted by sz;

* a link to a resource with ct Link-Format followed by a link to a
resource with ct CoRAL is probably not equivalent to a link with ct
CoRAL followed by a link with ct Link-Format if the server wants to
express that the CoRAL alternative should be preferred.

Klaus