Re: Section 4: LDPR/non-LDPR formal definitions

On 23 Mar 2013, at 02:20, Erik Wilde <dret@berkeley.edu> wrote:

> hello richard.
> 
> On 2013-03-22 9:01 , Richard Cyganiak wrote:
>> On 22 Mar 2013, at 15:37, Erik Wilde <dret@berkeley.edu> wrote:
>>> a resource's "nature" on the web is determined by its media type,
>> There is no such thing as "a resource's media type". Resources may have representations and these representations have a media type. A resource may have multiple representations, and clients may be able to interact with it through more than one media type.
> 
> you're of course correct, pardon my sloppy terminology here.
> 
>> Remember, the conversation is about whether we need to define a class of "non-LDPRs"; typing a resource that way would be an assertion that it does not conform to the criteria for an LDPR. I don't see a use case for this. Henry said "binary files are a use case". I disagreed with this; LDP can support "binary files" just fine. The class we need is simply "arbitrary HTTP-accessible resource", not "non-LDPR".
> 
> pretty much any media type i know has assumptions around where a link takes you. it may take you to a resource that supports the same media type, or not. while technically speaking, no media type should make assumptions about that, many applications will. when you follow an HTML img/@src link, you expect an image/* response, or you don't really know what to do. same for a/@href, you expect an HTML page, or you don't really know what to do.

Erik this has nothing to do with the media type, and all to do with the type of relation that the <img src="..">...</img>
relation is ( if you put on your RDF glasses you'll see relations everywhere ). If we were to write out the definition of
of the html:img/src relation we could do it like this:

html:img/@src a rdf:Property;
   rdfs:domain html:DocumentPart;
   rdfs:range html:Image .

So the reason you expect the object of an an img/@src to be an image is that this is part
of the semantics of the html:img/src relation . HTML could be written in xml, json, or lisp
it would still be the same: you would still expect the dereferencing of an image to return
you a media type that is something that will resolve to an image.

> 
> so what a media type might want to say is when links are "within" the media type (linkage between various resources exposed through that media type), and when they (potentially) "leave" the media type. this is helpful for clients to know, even though strictly speaking it would be best to never make any assumptions about this.
> 
> i do agree that it is not necessary to define "non-LDP" resources. what you find and what it means to you is determined by the link you follow, and the media type you get when you follow it. that's all clients need to know.
> 
> cheers,
> 
> dret.

Social Web Architect
http://bblfish.net/