RE: [xri] updated proposed XRI 1.1 ABNF

["Drummond Reed" <drummond.reed@cordance.net>]

For those who prefer a web page version of this ABNF, it is posted at:

http://xrixdi.idcommons.net/moin.cgi/Xri1dot1ProposedAbnfv2

=Drummond 

-----Original Message-----
From: Lindelsee, Mike [mailto:mlindels@visa.com] 
Sent: Monday, September 20, 2004 5:06 PM
To: xri@lists.oasis-open.org
Subject: [xri] updated proposed XRI 1.1 ABNF

After some lively discussion and email exchanges, the editor's group has an
updated proposed ABNF for XRI 1.1.  Apologies for having the discussions
off-list, but we felt that we were delving into detail that wouldn't be of
interest to the main body of the TC.  Now that we have a proposal (with a
couple of issues yet to iron out), we'd like to get everyone's feedback.

I believe that this ABNF captures all of the requirements that have been
agreed to for XRI 1.1 and has also been re-worked to have a parallel
structure and naming convention to the 2396bis ABNF.  The biggest change
that you are likely to notice is the relaxing of the syntax in the path part
of the XRI.  While we have defined a relatively strict syntax in the
authority part (since the resolution specification needs to define semantics
around that syntax), the XRI specification doesn't (and, IMHO, shouldn't)
say anything about the semantics in the path.  Basically, this puts the onus
of defining the semantics of the path part to applications that use XRI
(such as XDI).  This is exactly the same approach that 2396bis takes with
respect to the path part of URIs.  

Issues that have been raised with respect to this ABNF are:

  1.  Should the "//" be required in all XRI's or is it enough to require
the "//" only in URI normal form XRIs?  E.g., should we allow
"xri://@foo/bar" to also be written as "xri:@foo/bar"?

  2.  Should multiple xrefs be allowed in a sub-segment in the path?  E.g.,
"xri://@foo/(xref1)abc(xref2)".  The current ABNF allows this as I saw no
reason to restrict this sort of XRI.

  3.  Should the names for sub-segment productions in the ABNF have "sub" in
them? The relevant productions are named:

	xri-segment
	xri-segment-nz		;non-zero
	xri-segment-nc		;no colon
	xri-segment-nc-od	;no colon, optional delimiter
	xri-segment-nc-nx	;no colon, no xref

      If sub-segments are called out in the prodution names, the last 3
would be renamed to:

	xri-sub-segment-nc
	xri-sub-segment-nc-od
	xri-sub-segment-nc-nx

  4.  Are the productions that have alternatives hard to understand without
explicit grouping using parenthesis.  E.g., xri-hier-part.

Feedback on these issues and any other topics with respect to the ABNF are
welcome.  The editor's group hopes to come to agreement on the ABNF this
week and so would like as much feedback as we can get.

Thanks,

Mike

 <<draft-xri-v1-1-abnf-v3.txt>>