W3C home > Mailing lists > Public > semantic-web@w3.org > September 2014

Re: SPARQL Profile for PATCH [was Re: LDP Patch Format FPWD published]

From: Andy Seaborne <andy@apache.org>
Date: Sat, 20 Sep 2014 10:28:26 +0100
Message-ID: <541D48BA.409@apache.org>
To: semantic-web@w3.org


On 20/09/14 05:34, carmen wrote:
> - Implementers could simply plug in an existing general-purpose
> SPARQL engine
>
> if SPARQL is linked-data's SQL, it should have INSERT.. and does,
> no?

It does.  INSERT DATA ...

> particularly with SPARQL's smaller niche, implementing everything,
> from SPARQL-Update to property-paths is a lot to bite off. you're
> likely to end up with a rather large product and only a few full
> implementations.. maybe ORACLE, Franz,

I understood David to be suggesting it was subset of SPARQL, not the 
whole thing.

"""
a simple restricted set of operations were defined as a *profile* of 
SPARQL 1.1 Update operations.
"""

> something from Apache with a SAAS sponsor

Apache Jena does not have a sponsor, SAAS or not.

Other Apache projects may have people who contribute on paid time. 
Every project is different but none have "sponsors" in the sense of 
money directed to the project.  Apache projects don't (the overheads of 
managing the different money streams are daunting).

Most open source implementations (Apache or not) of RDF toolkits or 
SPARQL engines do not have sponsors.  Some, a few, do.

> many apps/CMSes/publishing-engines have emerged built on simple (and
> even cloud/webservice-hosted) key/value stores or
> distributed-filesystems full of .json files rather than what was en
> vogue a decade ago, fancy Python/Ruby "ORMs" abstracting away SQL
> (they'd already decided they didnt like it). digging around old
> Multics/CTSS docs there were query tools resembling SQL's modern
> syntax 40 years ago and given all the installs it's likely here to
> stay for at least another century. so if you're a fan i wouldn't
> worry..
>
> this in the spirit of minimum-viable-product and more flexibility in
> implementation choices (your LD-PATCH might affect a runtime
> graph-store, a collection of files, or you might even save the
> patches themselves and 'replay' them starting at snapshots like some
> of the more exotic (darcs/bzr/mercurial) versioning-systems)
>
> there's precedent, widespread-usage, and longevity of KISS-principle
> patching-means PATCH(1)                    General Commands Manual
> PATCH(1)

A contribution to that style is:

http://afs.github.io/rdf-patch/

which isn't even a "language".  It's a diff format.

(apologies for the advertising!)

> alexandre says "the editors challenged themselves for not adding
> higher-level features"

LDP Patch is focused on the needs identified by LDP and the data forms 
likely to occur there.

YMMV.

	Andy
Received on Saturday, 20 September 2014 09:28:56 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:53 UTC