Dynamically assign identifiers - UMD response

On Mon, Jul 11, 2005 at 12:41:04PM -0400, Kendall Clark wrote:
> 
> On Mon, Jul 11, 2005 at 11:25:16AM -0500, Dan Connolly wrote:
> 
> > ACTION: KendallC to ask Bijan to consider implications of answering
> > bNode bindings with created URIs
> 
> Consensus on our end seemed to be "eek, don't do that".

Ultimately, if we are to make bNodes addressable, we need to associate
the stable bNode with some symbol in rq23. (Re-using bNodes directly
changes the language.) Andy has proposed (eliding issues of how the
server tells the client which bNodes are addressable):

  - in the protocol -
    Protocol:
        &var:x=bNode:xyz
  - in the QL -
    SPARQL Function Extension:
        FILTER (ext:bNodeLabel(?x, "xyz"))
    Dynamically assign identifiers:
        { tag:example.com,2005:xyz foaf:homepage ?hp }
    Split the label space of bNodes:
        { _!:xyz foaf:homepage ?hp }

If Bijan et al were concearned with making bNodes addressable as URIs,
are the other forms addressing problems?

For instance, DL restricts triples involving anonymous classes.
Mapping those to URIs could allow someone to make arbitrary assertions
about them (or, at least, your view of them). Does that change if the
addressing mechanism is in a FILTER, protocol association, or special
_!:xyz node?
-- 
-eric

office: +81.466.49.1170 W3C, Keio Research Institute at SFC,
                        Shonan Fujisawa Campus, Keio University,
                        5322 Endo, Fujisawa, Kanagawa 252-8520
                        JAPAN
        +1.617.258.5741 NE43-344, MIT, Cambridge, MA 02144 USA
cell:   +81.90.6533.3882

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.