RE: Call for consensus on defining IRI template expansion (ISSUE-30)

On 17 Jul 2014 at 11:55, Ruben Verborgh wrote:
> On 17 Jul 2014 at 11:12, Tomasz Pluskiewicz wrote:
>> On Thu, Jul 17, 2014 at 10:29 AM, Ruben Verborgh wrote:
>>> As a client, I need to know how to use the URI template
>>> before I can make an API request. Seems necessarily in scope to me.
>>
>> Necessary to make a request, yes. But it's not something tied to
>> descrpition of a hypermedia API strictly speaking.
> 
> But it is tied to the description of a hypermedia control.
> Hydra on the one hand describes controls, and on the other hand APIs.

Think of an HTML form using GET. In principle it is the same thing as a
Hydra IriTemplate. The HTTP specification defines in detail how to turn the
form values into a URL query string. We need to do the same to ensure that a
client sends the information in the right form for the server to understand
it. 


>>>> The value could then simply be a turtle-formatted RDF object. Even with
>>>> the angle brackets to be consistent.
>>> 
>>> And also with escaping? 
>> 
>> For simplicity's sake yes. Though I'm not strongly in favor of angle
>> brackets and will certainly not miss them here. Especially with the
>> assumption that only absolute URIs would be allowed.

I think we should try to avoid the need to handle all the complexity that
Turtle escaping introduces. I don't want to introduce a dependency on Turtle
and I don't want developers to have to deal with it (or use a Turtle
library). The proposed expansion mechanism allows makes it trivial to decode
values: a simple regex or a couple of lines of code are more than enough.


>>>> Though I would not allow prefixed URIs.
>>> 
>>> Would be difficult indeed if no prefixes are agreed upon
>>> (and it's probably better not to).

+1


--
Markus Lanthaler
@markuslanthaler

Received on Monday, 21 July 2014 18:50:36 UTC