Re: [xacml] another small time/date issue

[bill parducci <bill.parducci@overxeer.com>]

unless i read this incorrectly it seems that these are not mutually 
exclusive positions: a request goes to a single PDP whereby information 
may be added/enhanced for downstream decision requests. in this 
scenario, attributes may be change in the life span of an aggregate of 
decisions, but they remain 'explicit' (static) for each decision context.

true, this may lend itself to unanticipated decisions in the macro 
sense, but each localized decison--that which i believe XACML is 
intended to cover--will resolve deterministically.

b

Daniel Engovatov wrote:
> It may.  But do we really want to specify exactly how it is done without
> unduly restricting architecture of implementations?  
> I thought it would be beneficial for an authorization language standard
> to stay out of data flow management and operational behavior as much as
> possible.  We specify how to deal with attributes that are explicitly
> provided in a request, but anything concerning PIP sources of
> information was intentionally left undefined, as far as I remember.  
> 
> Daniel.
> 
> -----Original Message-----
> From: Anne Anderson - Sun Microsystems [mailto:Anne.Anderson@sun.com] 
> Sent: Tuesday, August 12, 2003 5:36 PM
> To: Daniel Engovatov; Seth Proctor
> Cc: xacml@lists.oasis-open.org; Polar Humenn
> Subject: RE: [xacml] another small time/date issue
> 
> Won't the initial Request go to a single PDP?  And that PDP might
> invoke others to evaluate sub-policies?
> 
> If that is the case, then the initial PDP could add its concept of
> "current-time/date/dateTime" to the Request context that it sends
> to any other PDP for subordinate evaluation.
> 
> -Anne
> 
> 
> 
> You may leave a Technical Committee at any time by visiting http://www.oasis-open.org/apps/org/workgroup/xacml/members/leave_workgroup.php
>