Re: ProtocolDataModel issues

[Greg Stein <gstein@xxxxxxxxxx>]

On Sat, Sep 11, 2004 at 01:41:52PM -0400, Robert Sayre wrote:
> Danny Ayers wrote:
> >On Sat, 11 Sep 2004 09:34:41 -0400, Robert Sayre <mint@xxxxxxxxxxxxxxx> 
> >wrote:
> >
> >>Note that this data model is distinct from that of an Atom feed.
> >
> >Distinct, ok, but why different? Shouldn't the feed model act as a
> >starting point?
> 
> Well, sure. I was responding to comments that this work was "already 
> done." Think of this exercise as a model for the resources the server 
> must make available. There's more than just Atom format stuff there. For 
> example, there's the a lot of functionality that would typically go in 
> an admin screen in MT, Wordpress, etc. Adding/Removing users, adding 
> categories, stuff like that. Make sense?

I think it makes sense. And I also think that we need to be careful to
define some limits around what the spec will cover and then be able to say
"go use some other standard to deal with the rest of the stuff."
Personally, I'd like to see the resource types and properties defined for
all the things we need to manipulate, then just defer to WebDAV to do
that. And use its extensible property model for all the vendor-specific
stuff that people want.

> >>http://www.intertwingly.net/wiki/pie/ProtocolDataModel
> >
> >I'm also a little puzzled by, e.g. Post Page:  "A formatted display of
> >a single entry.".  -  the model part is the Entry, surely the
> >presentation should be dealt with separately?
> 
> The HTML representations are objects the model could take into account.

A URL points to a resource. Every resource should have a listing in our
data model. In this case, a post page is a typical item in a blog setup,
so I think it is important to have in the model. There are also
relationships that are defined between that page and others in a blog.

The important part is that the system is not necessarily a pure MVC, and
it may not be described that way. The model is implied by the rules of
HTTP, and its notions of resources and methods on those resources.

> >>Templates--
> >>Greg Stein: "how do we describe which syntax the template is using?"
> >>Could we use MIME? Vendors could register their template languages under
> >>vnd. or prs.

Ah. That sounds good.

> >Lost me. Could you please explain how the syntax of templates relates
> >to the data model.
> 
> The protocol must provide capabilities for modification of templates 
> (see charter). One of the properties of a template is its 
> syntax/language. One way of identifying the "language" of a template is 
> to give it a MIME type.

There are a number of reasons why you may want to know the syntax of the
template. Imagine you have Mother-Of-All-Blog-Tools and it knows how to
author 10 different, common template syntaxes. How does it know which of
those is in use? Thus, the need for a property.

Cheers,
-g