Hello Aldo,

On Mon, 2010-11-08 at 15:40 -0300, Aldo Bucchi wrote:
> Hi Ivan,
> 
> On Sat, Oct 23, 2010 at 11:40 AM, Ivan Mikhailov
> > I will implement batches of SPARUL statements.
> > OTOH it is possible to create stored procedures on your server, with any
> > logic inside, and call them from SPARQL. That's not a portable solution
> > but SPARUL with transactions is not a common thing anyway.
> 
> I politely disagree ;)
> I am currently aware of two systems in production that access Virtuoso
> only via HTTP ( read/write ).
> 
> So, in my world view, this is a pretty common usecase.

You are right, that it is common with Virtuoso, but it is questionable
if we look at the whole SPARQL society. Unfortunately not all users
migrated to Virtuoso, so they affect SPARUL statistics :)

I can't name it "undocumented feature", because both endpoint and SPARUL
are documented, but nevertheless it is neither covered by existing
SPARQL spec nor backed by mature transaction control (like setting
isolation levels, error handlers, standardized signals for unhandled
cases and complete diagnostics). The feature works and similar features
exist in other systems but they have many mismatching implementation
details. When it works, it works similarly on many different servers.
When something go wrong, it go different wrong ways.

As a result, a Virtuoso-specific stored procedure would not make SPARUL
interoperability worse --- it's too bad anyway.

Best Regards,

Ivan Mikhailov
OpenLink Software
http://virtuoso.openlinksw.com



Reply via email to