OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

xri message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: RE: [xri] polyarchical -vs- hierarchal


Les (and Wil, since this also addresses your question):

 

First, per the message I just sent, IMHO no normalization rule between @ootao+west and @ootao*(+west) can satisfy the human-usability requirement.

 

Second, I’ll try to be precise about the context differences. Each of the six context symbols defines a context as follows:

 

LOCAL CONTEXT SYMBOLS (LCS)

 

* à reassignable identifier assigned in a local context, i.e., can only be interpreted as identifying a resource in the context of the authority assigning the identifier.

 

! à persistent identifier assigned in a local context, i.e., can only be interpreted as identifying a resource in the context of the authority assigning the identifier.

 

GLOBAL CONTEXT SYMBOLS (GCS)

 

= à identifies the global context of identifiers assigned by individuals

 

@ à identifies the global context of identifiers assigned by organizations

 

= à identifies the global context of generic identifiers (tags) assigned by general community consensus

 

$ à identifies the global context of identifiers assigned by standards bodies in specifications for the specific purpose of interoperable identification of resources.

 

Having defined these, here’s what I meant by contrasting hierarchical and polyarchical usage:

 

1) The two LCS characters are similar to dots in DNS syntax or the slashes in URI syntax – they provide for hierarchical delegation between identifiers whose context is always local, i.e., their parent. Thus they have no polyarchical usage.

 

2) The four GCS characters have no precedent in DNS or URI syntax. They too allow for hierarchical delegation, but they allow the use of identifiers that *have their own independent global context*, i.e., are absolute identifiers in their own right. That’s the polyarchical usage – the ability to take an identifier rooted in one context and reuse (cross-reference) it inside another context.

 

In XRI 2.0, there was only one way to take an XRI that exists in one global context – such as +west – and put it in another context, and that was using parenthetical cross-reference syntax that forced it to also be put into a local * or ! context, i.e., *(+west) or !(=west).

 

However with direct concatenation, this forcing is no longer necessary. You can now put one global XRI directly in the context of another global XRI -- @ootao+west, @cordance+west – without being forced to say that its interpretation is limited to a local context.

 

Essentially, that’s what normalizing +west into *(+west) or !(+west) is doing: taking an identifier whose purpose is to be interpreted in a global context and saying “the following identifier SHOULD NOT be interpreted outside the local context”.

 

By contrast, allowing direct concatenation of @ootao and +west into @ootao+west finally enables us to clearly express the intended usage of the vast majority of cross-references: “the following identifier (+west) SHOULD be interpreted outside of the local context”.

 

Hope this helps,

 

=Drummond (now going to bed ;-)

 


From: Chasen, Les [mailto:les.chasen@neustar.biz]
Sent: Wednesday, May 02, 2007 10:18 PM
To: Drummond Reed; xri@lists.oasis-open.org
Subject: RE: [xri] polyarchical -vs- hierarchal

 

You are still not distinguishing very well the difference between local and global subsegments.  The descriptions you are providing in both cases are local delegations in the context of a parent.  So as Steve keeps asking why doesn’t @ootao*(+west) satisfy the requirement?  If the meaning is different it is unclear how so?  I thought today you were saying that @ootao+west is not a hierarchy but a polyarchy. 

 

 

contact: =les

sip: =les/(+phone)

chat: =les/skype/chat

 

 


From: Drummond Reed [mailto:drummond.reed@cordance.net]
Sent: Thursday, May 03, 2007 12:48 AM
To: Chasen, Les; xri@lists.oasis-open.org
Subject: RE: [xri] polyarchical -vs- hierarchal

 

Les,

 

Actually, in XDI (both the ATI and RDF models), all link relationships are represented as delegations (either *! delegation, or =@+$ delegation) inside a single XRI segment. All hierarchical parent/child relationships are represented as forward-slashes (different segments).

 

When you create a link between two resources, you are naming the arc that connects the two nodes. So for @ootao, you first need a starting node, which is @ (the “arc from nowhere”), and then “ootao” is the link to the next node. The same goes for +west -- + is the starting node, and “west” is the link (arc) to the next node.

 

In @ootao+west, @ is the starting node, “ootao” is the link/arc to the second node, and “+west” is the link/arc to the third node. It is true that + may point “west” to a different node (and different XRD) that @ootao points “+west”. That’s the difference between their contexts. However the purpose of the + namespace is to provide common context (generic identifiers), so the idea is that both +west and @ootao+west point to resources that serve as THEIR representations of the generic subject “west”. In the + dictionary space, this representation should be a definition. In the @ootao space, it could be any resource that @ootao identifies (and wishes others to be able to generically identify) as “west”. Note that if using “+west” all by itself might be ambiguous, @ootao can do the same thing Wikipedia does and add more context (Wikipedia refers to this as “disambiguation” – see http://en.wikipedia.org/wiki/Wikipedia:Disambiguation for a great summary of this.) To do this, @ootao could create XRIs such as:

 

            @ootao+office+west

            @ootao+division+west

 

Hope this helps,

 

=Drummond

 


From: Chasen, Les [mailto:les.chasen@neustar.biz]
Sent: Wednesday, May 02, 2007 8:54 PM
To: xri@lists.oasis-open.org
Subject: [xri] polyarchical -vs- hierarchal

 

Drummond –

 

In a conversation today, you stated, I think, that we need global subsegments so that we can move from simply a hierarchical relationship to a polyarchical one.  I think this means that with the advent of global subsegs we now can move from one hierarchy to another one.   I think in XDI you call these things links.   I thought we support this today through the use of REFs in the XRDS.  Now, I suppose, that you are proposing that we move this notion into the syntax layer.

 

These examples of hierarchical XRIs truly show local delegations.

 

@ootao*west

@ootao*(+west)

 

These examples of polyarchical XRIs demonstrate a global relationship (link)

 

@ootao+west

@ootao+west*steve

 

Is this accurate?  If so, my main heartburn is that since +west in @ootao+west as currently proposed is in the context of @ootao.  This means it could be completely different then the view from +.  If we are really linking to another namespace, +west, we should get +’s view of west not @ootao.

 

Thoughts?

 

 

contact: =les

sip: =les/(+phone)           

chat: =les/skype/chat

 

 



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]