Re: [arc-dev] SPARQL 1.1 - call for suggested features
From: Will Daniels
Subject: Re: [arc-dev] SPARQL 1.1 - call for suggested features
Date: Wed, 04 Mar 2009 13:18:16 +0000
Hi Benji,
I'd pretty much agree with that order. The first 2 in your list are
basic essentials for querying - aggregate counts and sums are frequent
requirements (for me and I think most people) and it's too much to
expect every SPARQL client needing to do such things to code it
manually. Plus there's probably some potential with these for variations
to start cropping up if it's not standardised soon.
Further to that, LOAD, DELETE and INSERT really are a minimum set of
features to allow the SPARQL language to function as a basic abstraction
layer for semweb applications to be more portable between different stores.
Although I find LOAD by far the most convenient here, I'd take DELETE
and INSERT over LOAD for standardisation if I had to choose (LOAD seems
to be something that can more reasonably be considered in context for a
specific store).
Projectable expressions would be very nice, but it's not too much
trouble to handle this manually in the majority of cases so I think I
might place bnode referencability above that (not that I've had cause to
use either yet myself to be honest). The last two don't really concern
me at all right now.
In summary: 1 & 2 essential, 3 & 4 extremely important, 5 & 6 very nice
to have, 7 & 8 just "nice".
Cheers,
Will
Benjamin Nowack wrote:
> Heyup all,
>
> The new SPARQL WG just sent out a call for feature requests[1].
> The next SPARQL version (SPARQL 1.1) will only be extended with
> very few selected features, so I'd like to ask here which extensions
> you have used and which you think are most important. We could then
> send a ranked list of feature requests to the WG, maybe making their
> decision process a little bit easier. Deadline is next Tuesday!
>
> Here are mine (ordered by personal priority):
>
> * Aggregate functions (COUNT, SUM, etc)
> * GROUP BY (required for non-trivial aggregates)
> * LOAD
> * INSERT and DELETE
> * Projectable expressions [2]
> * bnode referencability
> * Service description when accessing an endpoint w/o a
> query parameter
> * a syntactical equivalent to SQL's "IN", to check a
> term against a list of values (via Andraz Tori)
>
> Thanks for your input,
> Benji
>
>
> [1] http://bit.ly/BUGSK
> [2] http://www.w3.org/2009/sparql/wiki/Feature:ProjectExpressions
>
> --
> Benjamin Nowack
> http://bnode.org/
> http://semsol.com/
>
>
