Re: Update

good afternoon;

> On 2015-10-12, at 10:50, Karima Rafes <karima.rafes@gmail.com> wrote:
> 
> [ … several variations in endpoint iri patterns …]

so long as these variations accord with the information provided in the respective service description, a test harness must accommodate them.
yes, discovering the service description itself is an issue.
in order to resolve it, the harness must permit the tested service to somehow declare its endpoint locations, from which the harness can then determine which language the respective location supports.

in order to accommodate services which do not provide descriptions, one could permit one as an aspect of the registration.

> 
> EXAMPLE : Allegrograph  (uses a different endpoint for a query and a
> update and the parameter HTTP for Update is query)

if a service fails to recognize an “update” argument, then it would be non-conformant.

> .[…]
> 
> I know also that the parameter output format is different between the
> triplestores... but I have not the time to test all the database
> alone.

why is this significant?
as long as a store observes the requests accept header and generates a conformant response for a given requested encoding in what way does this variation constitute a problem?


best regards, from berlin,
---
james anderson | james@dydra.com | http://dydra.com