On Thu, 21 Feb 2002, Carlisle Adams wrote:
> Hi Polar,
>
> >
----------
>
>
From: Polar Humenn[SMTP:]
>
> Sent: Thursday, February 21, 2002 4:17 PM
>
>
To: Carlisle Adams
>
>
Cc: ''
>
>
Subject: Re: [xacml] RE: policy model, part 1
> >
>
> On Tue, 19 Feb 2002, Carlisle Adams wrote:
> >
> >
> A RuleStatement contains the following items.
> > >
- a RuleCore, which is a triple ("subject", "action", "resource"),
> >
> although one or two of the components may be missing (meaning "any").
> >
>
> I would still like to see place holders for that information, such as
> > <AnySubject/>, <AnyAction/>, <AnyResource/>, so that it is explicit in
>
> what it means. You can lock in the positions in the syntax as well, which
>
> might lead to easier processing.
>
> This is fine, although the syntax is probably a little bit uglier (e.g., the
> <subjects> element now needs to be a choice of PredicateExpressionType and
> <AnySubject/>, rather than simply a PredicateExpressionType that may have
> zero predicates (minOccurs="0")).
Yeah, but although the schema may be a bit more complicated, who cares?
It's a schema, not the use of the actual language.
If you end up with minOccurs=0, then you have to allow for the following
cases hunting around for them:
nil
S
S,A
S,R
S,A,R
A
A,R
R
And that is only if they are ordered as such S,A,R.
Although, I'm not up on my XML schemas, I would think that you could
create a SubjectPredicateExpressionType that was a choice of a
PredicateExpresstionType and <AnySubject/>.
That way, you *know* it is a predicate about a subject (resource,
action) by position and compile indexing functions much more easily.
Wouldn't it?
-Polar
>
> But I can live with either.
>
> Carlisle.
>
>
