[cap-talk] Re: "capabilities" as data vs. as descriptors - public key capabilities

[Jed Donnelley]

At 05:55 PM 5/3/2004, Valerio Bellizzomi wrote:
>Hey you should take a look at LoCI (http://loci.cs.utk.edu/) which is a
>project related to PlanetLab (see at
>http://www.planet-lab.org/php/related.php --> "The Internet Backplane
>Protocol site at the University of Tennessee, Knoxville."), and in
>particular at ExNodes
>(http://loci.cs.utk.edu/modules.php?name=Content&pa=showpage&pid=4_5), but
>be sure to read the page 2, especially the section on portability at the
>end of the page.
>There is other documentation too, I am still reading it.
>
>val

Funny you mention it.  Speaking with Scott Atchley from UTK, who
works on the LoCI project and supports some of the software that we
run here at NERSC, was one of my motivations to reopen this issue
of network capabilities - that led into the whole capabilities as data
vs. as descriptors discussion.

That and then the later discussion with a colleague about getting funding
for some computer security related work (possibly at LBL related to Akenti:

http://www-itg.lbl.gov/Akenti/

).  It does seem to me that there is a lot of confusing of terminology,
discussion at cross purposes, etc. in this area.  I was hoping to tie some
of the most fundamental needs together with the network capability concept
that supports POLA (Principla Of Least Access) at a minimum (and perhaps
at a maximum) to help move forward an implementation that might be simple
enough and then backward compatible enough (hence my initial emphasis
on interfaces) be become real (read widespread).

By this point I'm not sure I'm not hurting more than I'm helping, so perhaps
I should do some reevaluation.

--Jed http://www.nersc.gov/~jed/