Re: Trying to close ISSUE-14

On Wed, Apr 3, 2013 at 11:14 AM, Wilde, Erik <Erik.Wilde@emc.com> wrote:
> hello nandana.
>
> On 2013-04-03 05:28 , "Nandana Mihindukulasooriya" <nmihindu@fi.upm.es>
> wrote:
>>My attempt was to see your example under this proposal but as both you
>>and Raul pointed out, I fell into the trap of mixing protocol data and
>>resource data. As this is protocol data, it has to be ldp:index and it
>>should only appear in a page of
>> a container but not in member's description.
>
> there really is no such trap. instead, mixing protocol data into your
> resource representations is one of the essential design steps in REST,
> when you design your hypermedia controls. because that's what REST
> requires you to do: create representations that are self-contained, i.e.
> they represent what the client needs as state at a given moment, and they
> also represent the navigation capabilities for the client to engage in
> further interactions. so don't worry about mixing, instead, see it as
> moving in the right design direction.
>
> cheers,
>
> dret.
>
>

In thinking about it more as well, I see it as that as well (adding
some protocol behavior into the representation).  If a client happens
to take the triples and store what it gets as-is with these ldp:index
triples, it then needs to understand that if it is mixing updates from
further fetches of resources and that ldp:index value changes, it is
then undefined for the first container/page representation.  I client
that stores these resources, may decide to strip any triples with
ldp:index as the predicate without concern.

The question is, if we go with Raul's original proposal, we may want
to provide appropriate guidance on usages such as this.


--
- Steve

Received on Wednesday, 3 April 2013 19:17:53 UTC