More detailed XDI graph diagram example

[Drummond Reed <drummond.reed@xdi.org>]

On the plane back from London yesterday I worked with the new graph diagramming technique to do a more detailed "best practices"version of the basic Alice & Bob diagram. A copy is below. The primary difference is that now both Alice and Bob now have immutable IDs that $ref their mutable IDs.

Interestingly, I found that it was actually quite hard to try to orient the graph around the outer graph start node being in the middle of the left hand border, as I had advocated, vs. putting it in the upper left hand corner, as Markus suggested. Here's why:

1. If the outer graph start node is in the middle of the LH border, you have to decide what statements (branches) go above it or below it—and to switch a statement from being one to the other is a hassle because you have to reattach the arc connector. Conversely, if the outer graph start node is in the upper LH corner, then all connectors are below it, and you can just slide them up and down as needed to make room for more statements.
2. If the outer graph start node is not in the upper LH corner, it becomes arbitrary what statements should be shown in what order on the page. But if the outer graph start node is in the upper LH corner, there is a natural order for the statements:
1. Statements about the outer graph start node.
2. Statements about other peer graph start nodes.
3. Statements about entities and their attributes, organized by authority collections (e.g., [=], [@], [*]).

Net net: I'm now convinced our convention should be to put the outer graph start node in the upper LH corner. It also ends out being more suggestive of the JSON serialization.

Note that I also employed another simple technique for handling long XDI arc labels while still keeping the graph compact: put the label at the start of the arc and then offset the next node down a level so the arc labels don't run into each other.

Reactions?