What is the "Other party" in Attribute Handling called?

[Rod Widdowson]

> > I like the multiple recipient context idea, something like
> > getRecipients() on the AttributeRecipientContext, to make it efficient
> > for the attribute resolver to resolve shared attributes. But that
> > seems like further stepping away from v2, and I'm not sure how far we
> > want to go in that direction.
> 
> I don't want to right now, and I figured just allowing for a more
inefficient
> loop would be good enough for a use case we don't even have yet.

My instinct is that without a use case this is probably a really bad idea.
We have just managed to discover a way to dig ourselves out of the pit of
having multiple principles and this would just reintroduce them - via
another mechanism.  Of course in this case we would not be populating them
and so the possibilities for 'confusion' are limited, but it feels like a
step too far.

As we discussed in Columbus, if later down the road we do decide that there
is a use case, then we can add a multi-context version of resolution context
without breaking the ABI by testing for the super-class where we need to.

R