Re: containers examples should end in /

On 14 Jan 2013, at 15:46, Henry Story <henry.story@bblfish.net> wrote:

> 
> On 14 Jan 2013, at 15:27, Andy Seaborne <andy.seaborne@epimorphics.com> wrote:
> 
>> 
>> 
>> On 14/01/13 13:54, Henry Story wrote:
>>> The argument is that instead of having containers
>>> with URLs such as
>>> 
>>>  http://example.org/container1
>>> 
>>> they should be
>>> 
>>>  http://example.org/container1/
>>> 
>> 
>> Having examples that are sensitive to details of the URI used to access them is risking confusion.  Either of those is a URI for the container resource.  Better to make the examples robust to access name variation.
> 
> That is part of the Turtle and RDF/XML specs, as well as of the URI spec I think since
> relative URI resolution is sensitive to these things.
> 
> Perhaps a section in the spec explaining these features would be useful.
> 
> But I don't see why one should ignore the way those standards were built.
> 
>> 
>> Using a prefixed name is also readable.
> 
> But there is no need for this, and that also creates confusion.
> The prefix system is just a shorthand for a full URI. It therefore
> does not have the same meaning as a relative URI.
> 
> <> means "This document" wherever that document was gotten from.
> The full URL for the same document refers to the same document, 
> but does not have the same meaning.
> 
> As mentioned before this has been very clearly been pointed out
> in philosophy of language in the case of indexicals terms such as
> "I" and "now".
> 
> For the philosophical background on this in terms of indexicals you may 
> look at Elizabeth Anscombe's famous 1975 article "The First Person"
> http://mind.ucsd.edu/syllabi/01-02/270/pwd01F270/anscombe.html
> And the literature on de-se attitudes.
> Chrisopher Peacocke's most recent contribution to the debate in his recent 4 set of talks
> at University College London "Self and Self-Representation"
> https://itunes.apple.com/gb/itunes-u/self-self-representation-video/id418892016
> 
> bring this up to date to the latest work in philosophy of logic.
> 
> These arguments are arguments to the irreducibility of the first person,
> and make exactly the point I am making concerning clarity, namely
> that you can acquiesce to
> 
>   A bear is behind Andy now.
> 
> and not come to the conclusion that
> 
>   A bear is behind me now.
> 
> If say you don't know that "Andy" refers to you.
> 
> To put that argument in terms of this protocol, you can read a 
> document such that 
> 
> <http://example.org/container1/> a Container.
> 
> and not understand that it is speaking about itself.
> 
> <> a Container .
> 
> is much clearer. The second one is necessarily speaking about the
> container, the first one requires some thinking on URIs to work 
> this out.
> 
>> 
>> @base can be used to provide the preferred URI form.

On thinking about this more closely, I'd say it depends. If the 
context of the HTTP request is made explicit then there
should be no @base. If the context is missing then the @base
is a good solution.

Nevertheless, since these resources should be serveable over
HTTP, and since the spec should show how to do it elegantly,
the argument as to that collections should be URIs that end
in "/" still holds.

> 
> The exact same argument applies there.
> 
> Rather than adding a @base, I think it would be better
> to make the context explicit by adding the minimal HTTP headers 
> of the response, which  would make the context clearer.
>  
> 
> Henry
> 
>> 
>> 
> 
> A short message from my sponsors: Vive la France!
> Social Web Architect
> http://bblfish.net/
> 

A short message from my sponsors: Vive la France!
Social Web Architect
http://bblfish.net/