RE: Allowing variableRepresentation also on IriTemplateMapping (was: variable representations edited for spec)

On 31 Dez 2014 at 17:44, Ruben Verborgh wrote:
>> I completely agree that it would be nice to do
>> so in some cases but does it change anything apart from making the
resulting
>> URLs longer/shorter?
>> 
>> Absolutely. I just don't see how mixing representation formats helps
clients
>> or servers or enables new functionality that can't be realized otherwise.
> 
> The use case I mentioned earlier would not be supported:
> 
> Extending the triple pattern fragments interface with a structured
free-text object search:
>    ?subject / ?predicate / ?object / ?freeText
> where the ?freeText field allows control characters (*, ?, .).
> Then the first 3 are IRIs/literals, the fourth is free-text.

I still think a datatype would be more appropriate for this use case than a
separate variable representation format.


> We cannot simply split this in two forms,
> because then we cannot combine a subject filter and a free-text filter.

Right


> The datatype-based solution might work (even though I dislike that
modeling),
> but then again we'd have to specify this is allowed for that field.
> This is probably best discussed in the freetext thread.

Yeah, that's probably better. Are there any other use cases you can think of
that would require different variable representations?


--
Markus Lanthaler
@markuslanthaler

Received on Sunday, 4 January 2015 20:59:19 UTC