[uf-rest] Roy Fielding on WebDAV and PROPs

[Breton Slivka]

I have to admit that I am now finding this conversation is slightly  
more over my head than I thought. (though it's becoming a lot clearer  
to me since my last response). I should probably quit while I'm still  
ahead. However, I would like to clarify a few things.

Even in the matrix capacity, isn't the semi colon still being used as  
a mechanism to access properties? If you use x,y, and scale  
parameters on a map URI, are you not accessing the properties at  
those coordinates for the map object? The only difference I see is in  
the number of dimensions. If you're looking for a key=value property  
within an object it's merely a slight shift in syntax, say    http:// 
server/object;prop=title;  where prop represents the axis, and title  
is a coordinate in that axis.

A list of properties can be just another property. indeed, according  
to RDF a statement about a resource is a resource in itself. I see  
your point about potential naming collisions when trying to implement  
on a site which already uses the semicolon to access some properties,  
but then if we were worried about naming collisions we wouldn't  
bother with microformats.

All I propose to be a "standard" (better word would be convention) is  
the name of the property which would contain a list of properties.  
This is not too extreme I think. That way instead of PROPFIND being a  
method which prevents resources from having URI's, it becomes a  
resource with its own URI.

So   GET http://server/object;proplist  becomes its replacement.

does that make sense?




On Apr 11, 2006, at 10:49 PM, Mark Nottingham wrote:

> For example, if you have an exisitng site using WebDAV, and you  
> want to change it to this convention, you'll have problems if you  
> have existing resources with semicolons in their URIs.
>
> That's why I suggested that a site-configurable convention would be  
> more useful.