Re: [xdi] XDI literals as subjects/objects of relational arcs

Markus: ironic but I was just typing up my final notes on this for the TC Notes document.

I too thought about the old "value context node" perspective today. I think the big difference (and what Joseph ultimately objected to) was having an actual separate context node (apart from the attribute node) to represent the value. We all ultimately agreed it was not needed. So the solution was to eliminate that node but still have an address for a literal node that can be expressed as an XDI subject.

The net effect was to make literal nodes addressable AS IF they had a separate value context node, but without the value context node.

I also agree with you that this has substantial value even though it is a change. And that—as with all things with Core—now is the time to nail it down.

Onward to complete all the rest of the finishing details to get Core to a Committee Specification Draft vote next Monday.

=Drummond

On Tue, Oct 13, 2015 at 12:34 AM, Markus Sabadello <markus.sabadello@xdi.org> wrote:

I thought about yesterday's ABNF discussion a bit more, and now I remembered the model we had last year, when we still had the "value context node".

See attached simple diagram just as a reminder.

In the old model, it was perfectly normal for the "value context node" to be subject or object of a relational arc. For example, you could have a $ref link between one value and another value, instead of having a $ref link between the entire attribute and another entire attribute.

The use cases for this I think were rather thin. So when Joseph proposed and we all eventually agreed with him to remove the value context node, we saw the simplification and didn't see much disadvantage.

Now with today's ABNF, we're basically back at a model very similar to what we had last year, just without the value context node. The main argument in favor of supporting literals as subjects/objects of relational arcs is that they are real nodes with real XDI addresses in the graph.

So even though 1. this impacts existing implementation work quite a bit, and 2. the XDI/RDF mapping will be difficult, this does make more sense to me now than it did yesterday, in light of the model we had last year.

Markus

P.S. Here's a version snapshot of XDI2 before removal of the value context node: https://server.xdi2.org/pre-no-value-node/

---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail. Follow this link to all your TCs in OASIS at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php

xdi message