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] Updated 2.1 ABNF and global subsegments


Les,

 

We didn’t “factor out” xrefs. The refactored ABNF I just posted this morning at…

 

            http://wiki.oasis-open.org/xri/XriCd02/XriAbnf2dot1

 

…doesn’t propose any change to the ABNF parse tree, only changes to the ABNF rule names so that we eliminate semantic confusion over what we mean by the term “cross-reference” or “xref” for short.

 

The key problem that several TC members besides yourself pointed out when reviewing the 2.1 ABNF is that in the 2.0 ABNF we had only one concrete ABNF rule (to use Steve’s terminology) that instantiated the abstract concept of a cross-reference. That was the “xref” rule.

 

In the 2.1 ABNF we have two concrete ABNF rules that instantiate this concept, and calling only one of them “xref” was thus causing confusion.

 

The semantic solution I hit upon was to call both of them “-refs” (which they are). What was formerly “xref” is proposed to be “encap-ref” (for “encapsulated reference”) and the other form is a “global-ref” (which is one of the two forms of XRI subsegments, the other being “local-ref”).

 

So now the spec can explain, in the text, what I believe is the single most important feature of XRI syntax – the abstract concept of cross-references – and then refer to the two ways in which it is concretely instantiated in the ABNF, as either a global-ref or an encap-ref.

 

=Drummond

 


From: Chasen, Les [mailto:les.chasen@neustar.biz]
Sent: Wednesday, March 07, 2007 2:44 PM
To: Drummond Reed; xri@lists.oasis-open.org
Subject: RE: [xri] Updated 2.1 ABNF and global subsegments

 

I am not sure it makes sense to completely factor out xrefs.  Cross references are a very well baked concept in the XRI specifications.  To remove it seems rather drastic and a large change.  If you really want to make parenthesis optional in a xref why can’t you just make them so with [“(“][“)”] or something.

 

contact: =les

sip: =les/(+phone)

chat: =les/skype/chat

 

 


From: Drummond Reed [mailto:drummond.reed@cordance.net]
Sent: Wednesday, March 07, 2007 4:37 PM
To: Chasen, Les; xri@lists.oasis-open.org
Subject: RE: [xri] Updated 2.1 ABNF and global subsegments

 

Good points/questions, Les. See [=Drummond] inline.

 


From: Chasen, Les [mailto:les.chasen@neustar.biz]
Sent: Wednesday, March 07, 2007 10:19 AM
To: Drummond Reed; xri@lists.oasis-open.org
Subject: RE: [xri] Updated 2.1 ABNF and global subsegments

 

A couple of note/comments:

 

* another change that this includes but does not spell out is that ! is no longer a GCS character.

 

[=Drummond] Good point. Standard Example #10 on http://wiki.oasis-open.org/xri/XriCd02/GlobalSubsegments shows what happens under this proposal – the former ! GCS character (which by the XRI 2.0 ABNF had to be followed by a second !, so effectively it was always !!) is now a subspace under another GCS character. For example, the XDI.org !! registry would become a subspace of the @ registry, i.e., !!1000 would become @!!1000.

 

* One thing that confuses me is that the second instance of a global-subseg is a xref without parenthesis but is not called a xref. 

 

[=Drummond] That’s a good point that was also brought up by Laurie and Marty on our XRI $ Dictionary calls. It’s purely a semantic issue, but an important one. I went back and studied the v3 proposal ABNF closely, and I came up with a solution I really like: remove the usage of “xref” altogether. See my next message to the list with a full explanation of the v4 I just posted at http://wiki.oasis-open.org/xri/XriCd02/XriAbnf2dot1.

 

* In this example =$d*2006-02-17=example how does that compare to =($d*2006-02-17=example) and =($d*2006-02-17)=example and =$d*2006-02-17(=example) and =($d*2006-02-17)(=example).  Do these all mean the same thing? 

 

[=Drummond] Marty and I went over this same question on our XRI $ Dictionary call yesterday. The first thing I need to point out is that, although we discussed not needing a delimiter before parentheses in the 2.1 ABNF, for good reasons Marty and I discovered, we do need to require one (either a GCS or LCS character). So the last two XRIs in your question should be =$d*2006-02-17*(=example) and =($d*2006-02-17)*(=example). 

 

[=Drummond] On the question of whether they all “mean the same thing”, there are two possible answers: a) yes, they represent the same resource because they *normalize to the same XRI*, or b) they MIGHT represent the same resource (i.e., be synonyms), but that can only be determined via resolution.

 

[=Drummond] For good reasons which we’ll talk about on tomorrow’s call, right now my answer would be the latter. But I plan to discuss this with Marty further before then (and if we come to solid conclusion, we’ll post it).

 

=Drummond

 

 


From: Drummond Reed [mailto:drummond.reed@cordance.net]
Sent: Tuesday, March 06, 2007 12:47 AM
To: xri@lists.oasis-open.org
Subject: [xri] Updated 2.1 ABNF and global subsegments

 

XRI TC Members and Observers:

 

An update to the proposed ABNF for XRI Syntax 2.1 has been posted to the XRI TC wiki at:

 

            http://wiki.oasis-open.org/xri/XriCd02/XriAbnf2dot1

 

This third draft incorporates feedback and suggestions from earlier drafts. In particular it now breaks the former xri-subseg (XRI subsegment) rule into two forms: global-subseg (global subsegment) and local-subseg (local subsegment). Global subsegments replace the former compact syntax proposal. A full explanation of global subsegments and how they would be treated by XRI resolution has been posted at:

 

            http://wiki.oasis-open.org/xri/XriCd02/GlobalSubsegments

 

Note that the latter page still has a few TO-DOs at the end which I’ll fill in tomorrow. But I urge you to review both the proposed ABNF and the global subsegment proposal in preparation for this Thursday’s call.

 

Also, in a call I had with Les and Wil and Trung today, Wil took the action item to review the ABNF from the standpoint of the OpenXRI parser implementation. So he may come back with suggestions about how rule names might be refactored to best support implementation.

 

Please do send any feedback/comments directly to the list, as this is the last major outstanding issue for XRI Syntax 2.1, so we want to close on this and begin drafting as soon as we can to be ready for final review at the OASIS Symposium starting April 16.

 

=Drummond

 

 

 

           



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