OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

xdi message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [xdi] Fwd: Attribute and singleton are properties of a node, not of an arc

Really glad it makes sense to you guys.

After sleeping on the questions I posed at the end, I'm also clear that in XDI Display Format &^ need not be considered prefixes or suffixes of the arcs at all:
  • this keeps the graph model primary, and the serialization close to the graph model
  • avoids having to decide whether the characters are suffixes or prefixes, which avoids raising the question of whether this calls for switching the 6 context symbols to suffixes
  • nodes carrying something concrete are more visualizable, and discourages conflating arc labels with node labels. 
(Of course, in implementation code, a Java XDISubsegment class (or any class) could store them as class members if convenient for program logic.)

On Apr 15, 2013, at 11:15 AM, "Chasen, Les" <les.chasen@neustar.biz> wrote:

I like it.  Not only does the justification make perfect sense to me it also "looks" better.

From: Markus Sabadello <markus.sabadello@xdi.org>
Date: Monday, April 15, 2013 1:25 PM
To: Joseph Boyle <planetwork@josephboyle.net>
Cc: OASIS - XDI TC <xdi@lists.oasis-open.org>
Subject: Re: [xdi] Fwd: Attribute and singleton are properties of a node, not of an arc

I really like this analysis..

- The context symbols are really part of the arc's identifier, not of any of the involved nodes.
- & and ^ are constraints on the node that follows the arc, therefore these characters should be at the end of the arc's identifier.

I have to admit I have gotten used to having all the symbols in front of the arc's identifier, so at first glance =markus+email&^ looks a bit alien to me, but the logic behind it seems very clean. What do others think?


On Mon, Apr 15, 2013 at 4:35 PM, Joseph Boyle <planetwork@josephboyle.net> wrote:

I've had all along a strong intuition that the attribute symbol & and singleton symbol ^ should go after the name they have been preceding.

After examining the graph for this segment, I'm now clear that +markus+&^email needs to be +markus+email&^.

Let's graph +markus+&^email keeping mind that subsegments label arcs, and implicitly there is a node between each arc in the path.

           arc                  arc
node --------------> node ---------------> node
         +markus         &^    +email

The attribute symbol & signifies that the LAST node is capable of originating a value arc.
The singleton symbol ^ signifies that the LAST node can originate a value arc to "the" email for Markus.

and ^ are constraints on a NODE, not on a particular arc.

+markus+&^email puts &^ in the position of the middle node, which they don't apply to.
+markus+email&^ puts &^ in the position of the last node, which they do apply to:

          arc                  arc
node --------------> node ---------------> node
        +markus               +email     &^

So does &^ "really" belong with the preceding subsegment (as a suffix), or with the following subsegment (as a prefix), or neither (just a property of the node)?
Each of those three choices is an arbitrary convention equally compatible with &^ located between the arc in and the arc out of the node it applies to.

How about the six context symbols? They are really part of the arc's identifier, not the arc's source node or target node, so whether they precede or follow the name is arbitrary.  Preceding the name is what we are used to; following the name would be consistent with considering the attribute and singleton symbols to be suffixes.

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]