W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2011

Re: DELETE templates and blank nodes.

From: Lee Feigenbaum <lee@thefigtrees.net>
Date: Thu, 03 Mar 2011 11:33:32 -0500
Message-ID: <4D6FC2DC.4080501@thefigtrees.net>
To: Andy Seaborne <andy.seaborne@epimorphics.com>
CC: SPARQL Working Group <public-rdf-dawg@w3.org>
Thanks very much for the summary. Based on this and based on some IRC 
discussions, at Tuesday's meeting we will:

PROPOSED: SPARQL 1.1 Update forbids blank nodes in DELETE templates

While I realize this isn't everyone's first choice, it seems like the 
path of least resistance at this point in time. It also keeps the door 
open for an interpretation of bnodes in DELETE templates in the future.

Please let us know in advance of Tuesday's teleconference if you have 
significant concerns with this proposal.

thanks,
Lee

On 3/3/2011 10:35 AM, Andy Seaborne wrote:
> What are people's current thoughts about bnodes and DELETE templates?
>
> The only UC seems to be deleting fixed length lists. Are there any others?
>
>
> DELETE { ?x :p (1 2 3) }
> WHERE { ..find ?x .. }
>
> rather than something like (this should works on arbitrary length lists).
>
> ---------------
> DELETE
> { ?z rdf:first ?head ; rdf:rest ?tail . }
> WHERE {
> .. find ?x ..
> ?x :p ?list .
> ?list rdf:rest* ?z .
> ?z rdf:first ?head ;
> rdf:rest ?tail .
> } ;
>
> # Delete the triples that connect the lists.
> DELETE WHERE { ?x :p ?y . }
> ---------------
>
> Not ideal, especially as it does not combine easily,
> adding
> DELETE { ?x :p (1 2 3) .
> ?x rdf:type :XType . }
> WHERE { ..find ?x .. }
>
> so there is a risk "find ?x" will need to be repeated.
>
>
> My preference is currently to make a syntax restriction that bnodes
> can't be in DELETE templates (ditto DELETE WHERE) so a future WG can
> relax that and decide.
>
> That's
> "Option4: is to forbid bnodes in DELETE"
> from the last telecon.
>
> where "forbid" is make it a syntax error, done by adding an addition
> note to the grammar, rather than making the grammar have the productions
> for a triple block without bnode possibilities. This is how INSERT DATA
> vs INSERT is done.
>
> ==== Strawpoll summaries:
>
> The strawpool this time and last time hasn't been conclusive:
>
>
> Option 1: Bnode match all resources in the graph (essentially rewriting
> 1 in the mails)
>
> Option 2 : bnodes match whatever they match if treated as
> "modify_template copied to body"
>
> Option 3: treat bnodes as in CONSTRUCT/INSERT, i.e. as new bnodes...
> would mean they don't match anything.
>
> Option 4:is to forbid bnodes in DELETE
>
> Telecon: March 1:
>
> [bglimm] 3 or 4
> [pgearon] option 4 (followed by 3)
> [kasei] 0
> [sandro] (sorry, undecided.)
> [AndyS] no opinion - need to see details.
> [OlivierCorby] don't know yet
> [AxelPolleres] prefer 1 over 2 , but can live with 3/4
>
> and the time before: Telecon Feb 22:
>
> STRAWPOLL: option 1, 2, or 3? ←
>
> <pgearon> 3
> <NickH> 3
> <bglimm> undecided
> <kasei> 0
> <sandro> undecided
> <MattPerry> 0
> <cbuilara> 0
> <AxelPolleres> 1 or2 (but can live with 3)
>
>
> Andy
>
>
>
Received on Thursday, 3 March 2011 16:34:11 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:45 GMT