Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [higgins-dev] DigitalSubjectRef (was What is a relationship?) 
Jim Sermersheim wrote:
So basically, we need to make sure that a DSRef is able to be resolved to the right DS in the right Context (whatever "right" means). I guess the solution could go the way Tom is thinking (add another part to IDigitalSubjectRef), or we could just say that the ContextRef (URI) holds enough information to obtain the "same" context. We've already established (http://spwiki.editme.com/ContextRef) that two equal ContextRefs will produce "same" contexts. Is it fair to add that given two "same" cuid's and two "same" contexts we can produce two "same" DigitalSubjects? If so, problem solved. If not, then I think it's not right to say two same ContextRefs produce two same Contexts.
I think that's all we should require. To do more than this, IMO, places a significant burden on the Context implementation. A DigitalSubject that exists now might change or might not exist in the future. The Context should not be required to save all referenced versions of the DigitalSubject to preserve some view from the past. If the consumer cares about this, it can create a complete copy of the DigitalSubject in its own "permanent" local Context.
Analogy: If I login to my computer as root, I can see files that don't appear to exist if I login as gbyrd. But that doesn't mean there are two different file systems. It just mean that I have a different view of the contents of the file system. Using the same analogy, Unix does not guarantee that a file looks the same the next time I read it, even if it has the same name. And there's no guarantee that I'll be able to recover the old contents of that file if it changes -- if I care about the contents at a particular instant, then I'd better make a copy. Can a Context reuse the CUID of a DigitalSubject that has been removed? If so, then we can't assume that today's URI+CUID will refer to the same DigitalSubject as tomorrow's. But to say that a CUID can't be recycled seems too restrictive to me. 

...Greg
Note that "sameness" isn't obvious (as Tom started pointing out). The non-static nature of DigitalSubjects and the fact that we're not passing around authN/authZ materials and policy along with references will cause two same contexts to possible produce different views of the "same" DS (one Context may not even produce the DS at all!). Jim 

>>> "Tom Doman" <TDoman@xxxxxxxxxx> 8/15/06 9:40 AM >>>
OK, cool.  Another point I'm considering.  Any given context provider
may have policy\configuration\metadata (whatever we're calling it) that
governs the creation of Digital Subjects from that context.  For a
simple example, policy may constrain the attributes that are allowed to
be presented in any given Digital Subject produced from a given context.
Is any resulting Digital Subject only as long lived as the policy which
governed it's creation?  Certainly any given Digital Subject's
attribution may change at it's source and yet it remains the "same"
subject.  A change or removal of a governing policy will produce a
"modified" Digital Subject as well.

So, a long way to the question ...
Should we allow for the optional specification of a policy URI in
IDigitalSubjectRef?

This may be more important in the scen ario where a given Digital
Subject is the result of a join where what is joined from which sources
is governed by policy which may change.

I guess it comes down to this question ... As a consumer of IdAS, how
am I sure that a given DigitalSubjectRef I've remembered will produce
the "same" DigitalSubject each time?

Tom

>>> Greg Byrd <gbyrd@xxxxxxxx> 8/14/2006 11:31 AM >>>

It's both.  It is between DigitalSubjects, which may represent the same

entity or different entities.

We don't have the notion of a "persona" of a DigitalSubject.  A
DigitalSubject is defined solely by its Context.  "Home Tom" and "Work

Tom" are different DigitalSubjects, though they both represent the same

entity.  So a relationship between "Home Tom" and "Work Tom" is a
relationship between distinct DigitalSubjects, as you state in item (1)

below.   But the type of the re lationship could be "SameEntity", or
something similar that represents your item (2).

...Greg


Tom Doman wrote:
> What is the semantic intent of the "relationship" construct between
> IDigitalSubjects?  Is it:
>
> 1. Meant to convey some association to other distinct
IDigitalSubjects?
>  As in, Jim has a coworker relationship with Tom.
> 2. Meant to associate distinct personas of a single IDigitalSubject?

> As in, Tom has a "work" persona and a "home" persona.
> 3. Both?
>
> I ask for this clarification for a follow to the ContextRef+CUID
> discussion which is right in line with my e-mails about "Unique
> Identfication of Amalgamated Digital Subjects".  I see this
discussion
> as just another case where we have to work out unique identification
> issues.
>
> Thanks,
> Tom
> 
______________________________ _________________
higgins-dev mailing list
higgins-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/higgins-dev
_______________________________________________
higgins-dev mailing list
higgins-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/higgins-dev
------------------------------------------------------------------------

_______________________________________________
higgins-dev mailing list
higgins-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/higgins-dev

Back to the top