Separate endpoints for Queries and Updates

[Reto Bachmann-Gmür <re...@apache.org>]

Hi,

If I understand things correctly Fuseki currently provides two SPARQL
protocol services but each providing only one of the two sparql
protocol operations.

Are there plans to provide one service supporting both operations?

Cheers,
Reto

Re: Separate endpoints for Queries and Updates

[Andy Seaborne <an...@apache.org>]

On 15/07/13 17:45, Reto Bachmann-Gmür wrote:
> Hi,
>
> If I understand things correctly Fuseki currently provides two SPARQL
> protocol services but each providing only one of the two sparql
> protocol operations.
>
> Are there plans to provide one service supporting both operations?

Not that I'm aware of - SPARQL update and SPARQL query can be 
distinguished by MIME type or by query parameter so it could be added if 
a patch is provided.

But there is a good reason for them to be different - security.  One can 
be more protected than the other, e.g. public query and controlled 
access update.  Having different names makes it easy to control from the 
web container or from a reverse proxy.

There is an uber servlet that analyses any requests and dispatched  them 
to the right impl servlet.  Either it does already or can be made to do 
shared services on one endpoint.

The uber servlet is a single front end to all SPARQL services and it 
dispatches to the correct operation servlet.  It's not usually 
configured in - different names for different operations provides 
possible controlled access by the webcontainer and a single all purpose 
servlet defeats that.

	Andy

>
> Cheers,
> Reto
>