directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Jencks <>
Subject Re: Too much too intertwined (was Re: [Triplesec] [AuthZ] Comments on alex's proposed definitions)
Date Sun, 28 Oct 2007 08:12:33 GMT

On Oct 27, 2007, at 4:09 PM, Alex Karasulu wrote:

> David,
> I divided the topic into separate threads so we can discuss each  
> part separately
> in easy to read bite sized chunks.  This way most of us, mainly me,  
> could respond to
> comments within a reasonable amount of time.  Divide and conquer!

OK, I definitely should have talked about this before I commented on  
your proposed definitions.  I think we should keep all the terms in a  
proposed model in one email thread (e.g as I changed your posts to)  
since the terms are very dependent on one another and unless all  
parts of a proposed change are in one email we will not be able to  
get a consistent view of such a proposed change.  That's why I  
combined them: I can't even keep your original model in my head  
unless its on one "piece of paper".  However if you insist I will  
repost my comments as replies to your original emails.  I think that  
will result in much more confusion than keeping all the terms together.

> These small chunks paraphrase concepts within NIST paper using  
> simple yet clear words
> that many of us can relate to based on our experiences with  
> authorization without being
> mathematicians.

OK, but as i tried to make clear in my comments I do not find all  
your definitions clear and they are not the same as the model in the  
NIST paper.  I believe you suggested we should make our terms and  
definitions as clear as possible so I think my pointing out where I  
find your language unclear is entirely appropriate.
> We can elevate the conversation to that level later however we will  
> loose some people in
> the process.  The idea is to engage as many people as possible with  
> simple clear descriptions.
> Because we don't need to define things as a specification does to  
> have users give us good ideas
> based on their experiences.  We just need to use clear language.   
> That was the point, not to
> reinvent the NIST terminology, but to state them in the IT vernacular.
> There will be time for pulling out the material in the NIST paper  
> and discussing it's points
> verbatim but first we need to discuss and identify the problem in  
> clear language without using
> complex vocabulary on ideas that are mixed together across an email  
> taking 5 pages.
> I don't think you considered why I initially used this format.  
> Perhaps you may consider
> breaking down your thoughts into smaller pieces?  Maybe you can  
> reply to my previous
> posts instead of derailing those threads?

If you think that it is better to have shorter emails and no easy way  
to construct a proposed modified model after comments to many of  
those shorter emails, I'll be happy to take my comments apart into  
the individual emails rather than the verbatim aggregation I used.

I'd like to mention why I like the NIST model:  it's really easy for  
me to figure out what it means.  I can easily see how to implement  
the data model in java, in a relational database, or even despite my  
relative lack of familiarity with ldap, in an ldap schema.  When I  
think of operations involving the model, whether it be deciding if a  
user can do something or changing their permissions or examining  
permissions, its really easy to see how to do it.  When I look at  
stuff thats not directly from the NIST model everything gets  
muddier.  This applies to my descriptions of scope and denied  
permissions and roles, and many of your proposed definitions.  I feel  
like if we don't start with the NIST model we will be wasting time  
reinventing the wheel.
> Alex

View raw message