Re: LET/assignment

Comments below.

On Mon, Jun 28, 2010 at 10:01 AM, Lee Feigenbaum <lee@thefigtrees.net> wrote:
> I'd like to start another -- hopefully final -- discussion on the topic of
> whether or not we want to include the assignment feature
> (http://www.w3.org/2009/sparql/wiki/Feature:Assignment) in SPARQL 1.1.
>
> The brief history here, as best I can remember/reconstruct it, is:
>
> 1. Proposed as feature during WG's original requirements gathering phase
>
> 2. Received significant support but missed the cut for what we chose to work
> on (http://plugin.org.uk/misc/votes2.svg)
>
> 3. Raised on the -comments list by Holger Knublauch of TopQuadrant in late
> October and a few dys later again by Jeremy Carroll of TopQuadrant
> (http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2009Oct/0003.html
> followed by
> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2009Nov/0000.html)
>
> 4. Assignment/LET was discussed, at least informally, at F2F2. Other
> keywords were suggested (e.g. BIND) as alternatives to LET. No consensus was
> reached on whether to include a dedicated syntax for assignment.
>
> 5. We discussed these comments internally in a thread starting near
> http://lists.w3.org/Archives/Public/public-rdf-dawg/2009OctDec/0392.html .
>
> 6. Andy provided some details on how the semantics of LET and SELECT
> expressions relate in late November:
> http://lists.w3.org/Archives/Public/public-rdf-dawg/2009OctDec/0540.html
>
> 7. We were asked about this feature by a TopQuadrant user (or employee, I
> actually don't remember which) at the SPARQL 1.1 panel at SemTech.
>
> I believe in the intervening time, there are implementations other than Open
> Anzo (Glitter) and ARQ which include LET. Is this true?

Mulgara has it. As Andy pointed out, it was trivial to do.

> I believe that Andy feels that this construct would be a syntactic addition
> that directly invokes the current draft's Extend operation:
> http://www.w3.org/2009/sparql/docs/query-1.1/rq25.xml#defn_extend
>
> Based on this history, does anyone have feelings for or against including
> this work? I'd like to discuss over email and then probably in next week's
> teleconference. I'm hoping to resolve the issue (I've created ISSUE-57 to
> track this) relatively soon and move on.

I'd very much like it in there. The sub-select/project approach is
syntactically painful, and a repeat of the mistake of not having a
MINUS operation. Sure it can be done without the extra grammar, but
it's a common user requirement, and we don't want them to have to jump
through hoops to do a simple, and reasonably common operation.

Regards,
Paul Gearon

Received on Tuesday, 29 June 2010 00:14:43 UTC