W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > October to December 2010

Re: SPARQL Update 1.1 review part1

From: Andy Seaborne <andy.seaborne@epimorphics.com>
Date: Wed, 06 Oct 2010 22:14:57 +0100
Message-ID: <4CACE6D1.5050908@epimorphics.com>
To: Axel Polleres <axel.polleres@deri.org>
CC: "Passant, Alexandre" <Alexandre.Passant@deri.org>, SPARQL Working Group <public-rdf-dawg@w3.org>, Paul Gearon <gearon@ieee.org>

On 06/10/10 20:21, Axel Polleres wrote:
>
>> > 33) Section 4.1.8 CLEAR
>> >
>> > I think "ALL NAMED" would be clearer than "NAMED"
>>
>> I'd rather keep NAMED, and I think that's explicit enough (also
>> defined in the formal model).
>
> hmmm, I find ALL NAMED really more intuitive, to be honest... shall we
> put that up for discussion in a separate mail?

ARQ implements "NAMED" and it's in the grammar.
http://www.sparql.org/update-validator.html

[44]  	GraphRefAll	  ::=  	GraphRef | 'DEFAULT' | 'NAMED' | 'ALL'

ALL NAMED risks confusion - dropping one word changes the meaning to 
something that causes more data to be lost.

DROP ALL NAMED ==>
DROP ALL # Oops.

>> > 24) Can someone help me when the USING syntax was agreed? It seems that, since we no longer have DELETE FROM/INTERST INTO (since we now use GRAPH patterns to indicate the affected graphs), there is no need anymore to avoid FROM and FROM NAMED, so I strongly suggest to switch back to FROM/FROM NAMED also in update queries.
>> >
>> > IMHO, USING just creates more confusion than it solves ... another new keyword, and not easy to explain to anyone why FROM and FROM NAMED don't work here the same way that they work in Query...
>>
>> IIRC, decision was let to editors and Paul made the change.
>> Is that a major issue for you, Axel ?
>
> Frankly, I don't like it, since as we removed DELETE FROM, it seems unnecessary to introduce a new keyword.
> would be good to mark with an editor's note:
>
>  "The group is currently discussing whether own keywords USING/USING NAMED are necessary, or whether FROM/FROM NAMED can just be used analogously to SPARQL1.1 Query"

[42]  	UsingClause	  ::=  	'USING' ( IRIref | 'NAMED' IRIref )

If it were:
DELETE {}
FROM <uri>
WHERE {}

it might be read as implying DELETE applies to the FROM.

We have discussed this in the WG and in the special TC.  I wasn't a 
supported of USING but FROM is, IMO worse.  Is there new information to 
justify reopening to decision to leave it to the editors?

On 30/09/10 23:10, Axel Polleres wrote:
 > [ WITH<uri>  ]
 > DELETE { modify_template [ modify_template ]* }
 > [ USING [NAMED]<uri>  ]*
 > WHERE GroupGraphPattern
 >
 > -->
 >
 > [ WITH<uri>  ]
 > DELETE [{ modify_template [ modify_template ]* }]
 > [ USING [NAMED]<uri>  ]*
 > WHERE GroupGraphPattern
 >

The fragments should agree with the grammar; no need to do this until 
everything is stable to avoid rework.

The grammar is written to be as helpful to users and implementers as 
much as possible. DELETE WHERE is a different operation (with different 
semantics - e.g. blank nodes).

Only the forms that take template+pattern and so have duplication take 
WITH.  Rule [39] Modify. DELETE WHERE does not take WITH.

Only the forms with an explicit pattern take USING/USING NAMED.

	Andy
Received on Wednesday, 6 October 2010 21:16:00 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:44 GMT