Loading ...
Sorry, an error occurred while loading the content.

Re: [rest-discuss] Jot, RNA and ReST vs MoST

Expand Messages
  • Bill de hÓra
    ... But you will need something in addition to the schema, unless that the names/structures in that schema are very finely specified, ie using a formal logic.
    Message 1 of 32 , Aug 6, 2003
    • 0 Attachment
      S. Alexander Jacobson wrote:
      > On Tue, 5 Aug 2003, Seairth Jacobs wrote:
      >>
      >>So you are saying that I should be able to have an uber-client that would
      >>know how to act with *any* resource merely be examing its schema?
      >
      >
      > I am saying that if YOU know the schema of the
      > resource that should be all you need to have have
      > a client interacts with it using HTTP. You
      > shouldn't need an API in addition to the schema.

      But you will need something in addition to the schema, unless that
      the names/structures in that schema are very finely specified, ie
      using a formal logic. At that point it would stop being a schema and
      become an ontology of the sort the semantic web people are
      advocating. Without that level of specification, you only have the
      API signatures, which are not adequate (any old OO salt knows this).
      Interoperation requires contracts to be grounded in behaviour aslo.

      I agree with Seairth. MoST to me, sounds like a content model that
      would be a compliment to REST, but not a replacement. Certainly
      there's only so far you can go with mimetypes and conneg.


      > I mean per URI out-of-band data.
      >
      > With MoST, given a URI, you can GET a resource and
      > know what the resulting content-type, schema, and
      > data actually mean.

      You might want to say what you mean by 'mean'. For me here, 'mean'
      means machine processable metadata. Otherwise it's not so different
      from what we have now. Unless MoST has a formal semantics / model
      theory, I'm in dispute with your assertion and even them I think
      there are a few people on the list who will disagree with you.


      > The meta-location header is an admission of the
      > failure of ReST to deliver the
      > visibility/interoperability it promises.
      > Meta-location is an admision that you have to
      > associate new interface semantics with each URI,
      > and that all ReST gives you is the guarantee that
      > the interface will use GET, POST, PUT, and DELETE
      > as verb *names* and that GET will be idempotent.
      > Thats pretty weak.

      Ok, I think I get where you're coming from. Have you heard of speech
      acts?

      Bill de hÓra
    • Mike Dierken
      ... Don t forget user-agent.
      Message 32 of 32 , Aug 11, 2003
      • 0 Attachment
        >GET without an Accept header implies fidelity).
        Don't forget user-agent.
      Your message has been successfully submitted and would be delivered to recipients shortly.