26 Sep
2013
26 Sep
'13
7:52 p.m.
On 9/26/13, Ludovic Dubost <ludovic(a)xwiki.com> wrote:
If I understand properly an autonomous agent is an
intelligent agent that
can act on behalf of a user.
This can be done with scheduler jobs, however only "Admins" can create
these jobs and decide on behalf of whom they run.
This might work. There's a small chance it wouldn't.
this looks more complex. I'm not sure what you
need exactly here.
Note that you can rewrite your own security model and plug it in XWiki
You'll need to also rewrite the security setting UI
This proposition looks extremely dubious to me. Do you know why there
aren't any capability security models in any filesystems? Because
everyone who's investigated them has come to the conclusion the only
clean way to do this requires kernel support.
This is what I read roughly a decade ago as the explanation for
Grasshopper, or maybe it was KeyKOS. It may not be true but I ended up
concluding the kernel had to be blown up anyways. And I haven't heard
of any capability security models implemented into only filesystems
since then. And the OODBs which might have capability models (and
would support them easily) are more like OSes.
It appears we're in the situation where I know exactly how a cap
system has to work in order to make it usable to human beings
(something the other implementors of cap systems aren't too bothered
with, which explains why they're a failure) but I don't know any of
your technology. Like zip, zilch, zero. And the reverse for your good
self.