Re: use cases

On 23/08/2012 20:54, Rushforth, Peter wrote:
> David C.: I still don't see where I'm putting attributes into a
> namespace where they don't have one to begin with.



well you are certainly doing that but I'd misread Liam's proposal. It
does support that (with the system needing to supply a system defined
prefix if you need the resulting model to be compatible with xml 1.0 +
namespaces.

The comment quoted above is a litte strange to me as you explicitly
called out that this is happening:

> tell the client that href means
> {http://example.com/namespaces/hypermedia}href

unprefixed attributes (in XML 1.0+NS) are always in no namespace.
You are using Liam's namespace proposal to interpret unprefixed
attributes as being in the hypermedia namespace. I queried whether that
was an extension to Liam's proposal but as Liam confirmed that is
supported by the draft. If your namespace definition document says that
href="zzz" should be interpreted as
{http://example.com/namespaces/hypermedia}href ="zzz" then it can be
parsed as such and if serialised as XML 1.0 would come out as
ns1:href="zzz" xmlns:ns1="http://example.com/namespaces/hypermedia"
which is OK. (although actually you'd get more traction and more
immediate implementation (if only partial) coverage if you used the
existing xlink namespace rather than inventing a new one.

an unobtrusive namespace document (or the recently re-proposed
architectual forrms:NG ) that said href attrbutes were xlink:href
would go a long way to meeting the major use cases for hypermedial
attributes.

David

Received on Thursday, 23 August 2012 21:25:00 UTC