Re: [xdi] Minutes: XDI TC Telecon Thursday 2011-12-15

[Michael Schwartz <mike@gluu.org>]

> 3) BOOTSTRAP PERMISSIONING ISSUE
>
> Mike explained that one issue the OpenXDI team has been debating is how to
> bootstrap the root link contract. Mike favors allowing self-referential
> link contracts implicitly. Drummond pointed out that while this may make
> sense for root nodes, it doesn't work for deeper nodes because there are
> too many examples where the XRI used to identify a node inside another
> context would not give the authority represented by that XRI permission to
> edit that branch of the graph.
>
> For example, if =drummond graph stores an =mike node (contextually
> producing the XDI address =drummond=mike), that does not mean =mike has
> permission to edit the branch at that node. That branch may simply be
> storing =drummond's notes about =mike, which only =drummond can edit (or
> give a link contract to edit).
>
> Drummond does favor a compact form of link contract that expresses a $all
> or $do$all permission for the authority identified by a context node.
>
> # ALL: Work on suggestions for how to express such a compact
> self-referential link contract.

To clarify, the example above, =mike would not be able to access 
=drummond=mike without a link contract. The question I was posing is if 
=drummond could edit =mike without a link contract. I think yes, because 
=drummond owns the graph, so he should have super-user priviledges to 
manage his own graph.

After discussing it, I think we should take Markus' suggestion and use a 
standard link contract for self-modification. How about $self$do ? This 
link contract would not have to exist, similar to the design for 
$public$do.

- Mike