Re: HRAC submeeting team meeting July 30, 1998: minutes

[Konstantin Beznosov <beznosov@baptisthealth.net>]

> I have a couple of questions about the minutes.
> 
> 1. Can the word "target" used in the list of Design Goals be replaced 
> by the words "ADO Client" defined in the subsequent "Rough Model and 
> Terminology in It" section with no change in meaning?

yes. It should have been replaced. I just did not propagate the change.

> 
> <  2. Tree-like hierarchy (acyclic directed graph [ADG]): each
> <  resource is uniquely identified by an identifier which has internal
> <  structure. The identifier structure represents the path to the
> <  resource. This resource space is similar to Unix directory tree
> <  organization, where a name of any file is a global path to that
> <  file beginning from root. An authorization rule allows to make
> <  decision on any node or leaf. Access to the resource is granted if
> <  access is granted to every node in the path to the resource. For
> <  example, in order to have access to resource identified by
> <  /BHSSF/SMH/patients/JohnSmith/bloodtests/August-02/test-3 access
> <  should be given to /BHSSF, to /BHSSF/SMH, to /BHSSF/SMH/patients,
> <  ..., to /BHSSF/SMH/patients/JohnSmith/bloodtests/August-02/test-3.
> <  The described logic is the same as in access control model of Unix
> <  file system.
> 
> 2. Is the proposed "tree-like hierarchy" restricted to a tree or can
> it be any acyclic directed graph (ADG), i.e. can there be more than
> one path to a node? For example, symbolic links provide multiple paths
> to a Unix file. All trees are ADGs but not all ADGs are trees. 
> Unfortunately, the term "hierarchy" is used for both things that are 
> arbitrary ADGs and things restricted to trees.

If multiple path to a node introduces extra complexity (design, implementation,
usage), then I think we want to do it only if it is necessary. Do you think it
is?

>    
> 3. The text implies that there is a single "root" for all resources.
> Is this the case, or can there be a forest of "tree-like hierarchies"?

The meeting participants did not discuss this issue.
I do not think it increases design, implementation or other complexity, unless
(as in Unix) there is reason to use relative (as opposed to absolute) paths. If
my assumption (no extra complexity) is correct then why not.

Will tree-like hierarchy without multiple paths suffice for resource space
organization?

Konstantin


----------------
Broadcast message to hrac-rfp from Konstantin Beznosov <beznosov@baptisthealth.net>.
Go to http://cadse.cs.fiu.edu/omg/hrac-rfp to browse the mail list archive.