[cap-talk] small notes re: waterken

[David Barbour]

On Fri, Mar 4, 2011 at 11:10 PM, Rob Meijer <capibara at xs4all.nl> wrote:
> I feel this construct, that still allows for C++ style destructor based
> RAII should be the way that object capability languages work.

I feel we should more broadly reconsider the whole notion of resource
acquisition and resource management from a security perspective. I've
said before that, rather than associating memory with 'regions', we
should instead be associating them with a 'purse'. I have a lot of
Nick Szabo's research in this area (cf.
http://szabo.best.vwh.net/scarce.html). But that research is still
immature. Even without jumping to a full market-based solution,
though, I feel we could avoid explicit 'acquire/release' patterns in
ocap systems. (They're fairly problematic for distributed programming
and orthogonal persistence in any case.)