[Osr-101] Neutron specs

[Thomas Comiotto]

Hi Andi
>
> I'm still not convinced. So you are absolutely positive that we would 
> never hit a usecase which could not be solved by simply providing a 
> different style?
>

No I am not. But we need to draw a clear line between when to declare 
multiple stylesheets vs. using/introducing a seperate container element 
that provides for a seperate "view configuration". We could drop 
support for multiple styles but that would be a pitty because it's nice 
and simple (it's also what people are used to when coming from html/css 
xml/xsl world). Moreover dropping multiple styles would force users to 
state redundant open uris in case that all they want to achieve is 
using an alternative style for a otherwise identical resource 
(including open uri).

Hmm. Let's support both, but switching styles first:)


> Uhhm, yeah. But with regard to a specified set of values: wouldn't 
> this again limit us with regard to extensibility? I feel that we can't 
> think about all the possible protocols people might want to use 
> Neutron with.

That's right.

>
> Although the introspection document is mainly served
> , the actual content might not. Think for example about a CMS which 
> serves its content for viewing via normal webpages and HTTP, but they 
> would like to use some proprietary protocol to deliver the content for 
> editing etc.
>
> As an easy example, they serve editable documents via a distributed 
> file system, e.g. AFS. So what we would end up with in the <open> url 
> would be a URI of scheme file. I admit that this example is a bit lame

No it's not - I was thinking about this one all the time..


> , because there you could simply leave "method" unspecified. But OTOH, 
> we then also had to take such a case into account when validating, 
> namely that "method" MAY be not be present.

How about ..,..?.. transport="PULL". No seriously, I now think an 
optional "method" attribute is just fine.


--
Bests
Thomas