Re: Do we need DELETE and CLEAR?

It's not been a support issue with the released update language 
documented in the submission.

Style-wise, verb-rich does not worry me.  The verbs should have a 
logical categorization, separating deleting triples and managing graphs say.

In one design, theer was "REMOVE" for "DELETE DATA" - that did not 
receive much enthusiasm.

On 11/03/2010 08:29, Steve Harris wrote:
> Observation about SPARQL Update syntax:
>
> SPARQL Update is quite verb-rich, making it (IMHO) harder to learn the
> syntax, and it seems to me that one of CLEAR and DELETE is redundant now:
>
> DELETE GRAPH <a>
> DELETE DEFAULT # see previous mail

I find that confusing usage - is that CLEAR or DROP?

You explain below it's CLEAR but reading it in natural language, I would 
have gone for a reading of get rid of the GRAPH (i.e. DROP).

It is useful to have an explicit "delete everything" operation (rather 
than say "DELETE WHERE { ?s ?p ?o }" both for simple implementations and 
also the application writer because it's quite a major thing to do.

CLEAR is an abbreviation of the abbreviated syntax (half :-).

c.f. SQL TRUNCATE.

Identifying the DEFAULT graph seems an interesting thing to do.

	Andy

> DELETE DATA { <s> <p> <o> . }
> DELETE WHERE { <s> <p> ?o . }
> DELETE { ?o <p> <s> . } WHERE { <s> <p> ?o . }
>
> The first one is equivalent to
> DELETE WHERE { GRAPH <a> { ?s ?p ?o . } }
> which is perhaps more evidence that two verbs don't make it clearer.
>
> We do still need DROP GRAPH, as that does something slightly different.
>
> - Steve
>

Received on Thursday, 11 March 2010 10:35:08 UTC