Summary of qname/bnode proposal from last week

This is intended to capture the ideas brain stormed in last week's meeting.

http://lists.w3.org/Archives/Public/public-rdf-in-xhtml-tf/2005Sep/0016.html

Problem statement.

  RDF/A looks good when we restrict ourselves to href and about 
attributes with URI values.
   This fails to address two requirements:
   a) use of compact URI abbreviation in values (required IPC)
   b) inclusion of bnodes (required RDF community)

We note that (a) could nearly but not quite be addressed by qnames since 
the syntactic restrictions on qnames would prohibit an identifier such 
as ipc:123-456-789, since the string after the colon is not an NCName, 
since it does not start with an NCNameStart character.

(a) is addressed by a compact URI which is roughly
     NCName ':' *
(b) is addressed by traditional bnode syntax (generalised a little)
     '_' ':' *

The problem if which attributes to use them with is addressed by 
extending the range of href and about to permit a new form of either (a) 
or (b) enclosed in '[' ']'

Thus

    href="[_:bnode]"

    xmlns:ipc="http://ipc.org/categories#"
    about="[ipc:123-456-789]"

would both be legal, the latter would be equivalent to
    about="http://ipc.org/categories#123-456-789"
using the concatenation approach familiar from RDF/XML and N3.

We were undecided as to whether to use the '[' ']' syntax, or an 
alternative ':' '' syntax. The two examples with the alternative syntax 
would be
    href=":_:bnode"

    xmlns:ipc="http://ipc.org/categories#"
    about=":ipc:123-456-789"

Both these proposed syntaxes are chosen to avoid strings that match the 
URI production of the URI RFC.

Hence legacy data, which does not use the new syntax, is fully 
interoperable with new systems. And new data, using the new syntax, is 
rejected as syntactically invalid by legacy systems (although such 
checks are mainly only done when a retrieval action is required).

Those present at the telecon informally agreed that this seemed to 
address our two hardest issues adequately. (Although no straw poll or 
vote was taken)

Jeremy

Received on Tuesday, 27 September 2005 13:22:22 UTC