Re: [tm-pubsubj-comment] ISSUE 1 - PS Doc Identification andauto-referencing

[Lars Marius Garshol <larsga@garshol.priv.no>]

I feel that this issue is stated in the wrong terms.

Isn't the issue really "how do we know which URI is the PS Doc
identifier?"

My immediate response to this is that I think we should make a note of
this issue AND LET IT BE FOR NOW! We need to figure out a lot more
about the PS Docs before we're ready to say anything about how to
figure out which URIs they are.

The current discussion assumes far too many things, and many of those
assumptions may be proven wrong before we're done, which means that
our resolution to this issue will most likely have to be redone later
on.

This discussion assumes that:

 a) self-reference is an acceptable way to identify published
    subjects,

 b) the PS Doc set is structured in a particular way and uses certain
    particular formats, and

 c) more things that I find difficult to express at the moment.

The real issue here is: what is the appropriate structure of a PS Doc
set? I feel that is what we really ought to discuss, and that all
these other issues should be left on ice for now. (BTW, starting
discussions of more than a few issues at the same time is not a good
idea. It means too many balls in the air at the same time, and a high
threshold for entering the discussion. (I know, because I'm finding it
very difficult to enter it right now.))

* Bernard (in issues.htm)
| 
| The PS Doc, 

What does it consist of?

| which is an addressable subject, 

Is it?

| is used as an auto-referencing PSI.

Is it?
 
| Should not the notion of auto-referencing PSI be defined in the
| introducing section or/and glossary?

Not before we know if we're going to use the concept or not.
 
| Proposal:
| 
| Introduce the distinction between addressable and
| non-addressable subjects identified by PSIs, leading two the
| definition of three classes of PSIs:
| 
| -- Class 1: The subject is non-addressable
| -- Class 2: The subject is addressable, but it is different of
|     the subject indicator itself
| -- Class 3: The subject is the subject indicator
|     (auto-referencing PSI)
| 
| PS Docs should belong to Class 3.

I think this is confusing. I would analyze the problem differently:

 - the PS Doc Set is a subject,

 - therefore we can create a topic that represents it,

 - the PS Doc Set will live at a well-defined URI,

 - therefore we can address it formally with either a subject address
   or a subject identifier (it doesn't matter which just yet),

 - every PS Doc set will be an instance of the class 'PS Doc set',

 - as we can have a topic for the doc set, we can also formally
   declare that the topic is an instance of the 'PS Doc set' class,

 - therefore we need a PSI for the notion 'PS Doc set'.

So my recommendation would be:

 - define a PSI for the 'PS Doc set' class,

 - any topic which claims to be an instance of that class represents a
   PS Doc set.

Whether the PS Doc set should be the subject or subject address of
that topic should be listed as an open issue.

I still feel that this is the wrong thing to be discussing, however.
We should concentrate on PS Doc sets and their structure, rather than
details like these.

* Steve Pepper
| 
| - how to differentiate between topics intended to be used as
|    PSIs and other topics

This should be listed as a separate issue.
 
| The solution (to which, by the way, I am by no means wedded) was to
| specify the subject identity of a PSI topic using a subject
| indicator reference to the topic itself. This approach, which I
| called this "self-reference", seemed to kill two birds with one
| stone.

Provided the PS Doc set is structured in a particular way. We should
first figure out whether it is or not.

| So I don't see the point of your proposal for distinguishing between
| three classes of PSIs.

Nor do I.

-- 
Lars Marius Garshol, Ontopian         <URL: http://www.ontopia.net >
ISO SC34/WG3, OASIS GeoLang TC        <URL: http://www.garshol.priv.no >