[cap-talk] [e-lang] Are Guards Ambient Authorities?
captalk at grant.x43.net
Thu Oct 8 07:21:28 PDT 2009
Rob Meijer wrote:
> Could you explain why guards would constitute any authority, either
> ambient or not?
(As Keven pointed out, I was wrong about the ambience.)
The integer guard constitutes the ability to turn an object into an
integer. I tend to conflate objects that confer non-local abilities
with authorities, but I could be misunderstanding what authorities
are, generally. By "non-local abilities" I mean things that can only
be done with objects other than those that are or were parameters.
> As a big fan of both generic programming and POLA related
> constructs I would further propose that most (if not all) patterns that
> rely on things like type guards for POLA at the expense of not being
> generic are simply not applying the proper patterns to the problem.
I agree, I think; my mental model of object-capability systems, until
now, has always been duck-typed without even any type-guards. Since
discovering type guards, I've been concerned about what happens when
you extend the principle to cover more types, even though many type
guards are obviously useful. Using them to the extent I'm worrying
about would be a bad pattern.
More information about the cap-talk