[zeromq-dev] Listing public resources in the XRAP Resource model

[Pieter Hintjens]

The ZeroMQ XRAP spec has nothing to do with RestTL, just to be clear.
Different generations of the same design over different transports.

To answer your question, I guess you'd fetch a URN like /music, and
it'd return all public objects. The goal is, indeed, to avoid magic
URNs except at the root level.

On Thu, Oct 15, 2015 at 12:18 PM, Tom Quarendon
<tom.quarendon at teamwpc.co.uk> wrote:
> I’ve a few questions about the resource model described in the XRAP rfc (and
> related RestTL), to clarify my understanding if that’s OK?
>
>
>
> First, a couple of statements. The following is true?
>
> The URNs of public resources are (can be) known, a client has (can have) a
> priori knowledge of them, and the client is allowed, intended, to construct
> those URNs. These URNs represent the “roots”, the “entry points” if you
> like. A client is supposed to know in advance what public URNs they want to
> access. However a client can also follow links to get to other public
> resources, it is not necessarily the case that all links point to private
> resources.
>
> Private resources, by definition, can only be discovered by following an
> href link from a public resource. However not all links would point to
> private resources.
>
>
>
> How do I list public resources? So in the example in the RFC, how would I
> get a list of the play lists? My suspicion is that you are not supposed to,
> that that’s not the intention. That is, I don’t think that the intention is
> that I do a GET on /music/playlist in the naive way. I suspect my question
> indicates a basic lack of understanding.
>
>
>
> So how would I get a list of the playlists? If I want to do that, is there
> some other public resource that I have to request instead? Not sure what
> that would be though. /music/collections/playlists? That wouldn’t work
> presumably because different resources under /music/collections would have
> different schemas.
>
>
>
> I think I understand the intention of the public/private division, in terms
> of forcing clients to follow links and not construct URNs, I’m just
> struggling to apply it to things like collections and child resources (where
> a child is existentially dependent on a parent).
>
>
>
> Clearly I *could* design my resource model in any way I choose and still use
> the XRAP protocol to access it. My suspicion is that the intention of the
> resource model is to lead you to a design that is “Richardson Maturity Model
> level 3” (i.e follow links, rather than construct URNs), rather than level
> 2, and I’m wanting to understand how to apply that to try and understand the
> advantages that it might bring.
>
>
>
> Any clarification would be excellent!
>
> Thanks.
>
>
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev
>