Re: PUT and extensions

Good question, one which needs to be sorted out
and well documented.

Some invariants I'd look for are

- If I put(U) then I can get (U) afterwards.

- I should be able to get anything I write converted on read using conneg
 
- It may be useful to have URIs for the specific RDF/XML form, etc.

I think the best way is to never use the ".rdf" over HTTP.

On 2012-04 -04, at 19:04, Henry Story wrote:

> Currently if I PUT a file with an extension such as .rdf or .ttl
> on a read-write-web server, should the server pay particular attention
> to those extensions? Or should it be agnostic about them? 

It should certainly never read anything into them.
It should look at the content-type to see what sort of a file is being PUT,
and then either store it with that metadata as it is, or if it 
is in a typical config which uses extensions, it should force the 
extension to 

> 
> I suppose I am asking how close an rdf read-write-web server is to a
> web-dav server? 

It is pretty close, with limited WEBDav facilities, plus the SPARQL/Update patch facility
of course.

> 
> If I PUT foaf.ttl
> should a GET foaf.rdf then succeed?
> what about a GET foaf HTTP/1.1 ?

I think that is bad practice, best to 
PUT foaf and GET foaf  with conneg translating on the fly, 
with the file on the server being foaf.ttl typically
so that other extension-based things in the same filesystem
will do the right thing.


> 
> Henry 
> 
> 	
> Social Web Architect
> http://bblfish.net/
> 
> 
> 

Received on Thursday, 5 April 2012 17:58:55 UTC