Re: URLs in Data Working Draft

On 15 July 2013 11:28, Jeni Tennison <jeni@jenitennison.com> wrote:

> Melvin,
>
> On 11 Jul 2013, at 12:13, Melvin Carvalho <melvincarvalho@gmail.com>
> wrote:
> > "When presented with a hash URL, such as
> http://photo.example.com/psd/12345#comment-67890 or
> http://photo.example.com/psd#me, applications can locate its content by
> resolving the base URL"
> >
> > My experience on the web is that it is difficult to explain to people
> the value of separating documents and data using the # value, and also of
> having multiple data items in a document.  For example you might have an
> identity for a user, and also, a public key for authentication and signing.
>  IMHO, publication of this document will make explanation that much harder.
>
> In what way?
>
> > I can picture a scenario of a divided web where documents are used as
> entities in wilful violation of TAG recommendations, and less interoperable
> with standards compliant material.
>
> Can we dig a bit more into what you mean by 'used as entities'? Do you
> mean that there are properties that are associated with the URIs that
> resolve to the documents? Can you give an example where this 'divided web'
> causes problems for applications consuming the data?
>

Hi Jeni, thanks for getting back!

As a background, I'm a grass roots developer that for several years I've
been following the specifications work at both the W3C, the IETF and
various group in the FOSS movement.  My dream is to see various projects
come together in an interoperable way to create a social web that is
federated across many nodes and heterogeneous projects.  Over the years
we've encountered a few challenges, let me try and explain how this is
relevant.

There appears to be a group that is well versed with awww (which I consider
myself to be still learning!), and in particular the trade-offs between
using fragment identifiers in web pages, wrt the data web.  However, that
many people outside the Linked Data community, imho, do not have a good
idea of the value of using fragment identifiers in documents.  And some
will not support that pattern at all.

Examples I can think off from the top of my head indieweb [1], tent [2] and
to an extent OpenID at least in some versions had the idea of 'your profile
is your identity'.

On the face of it, this may not appear to be an issue, as the web is all
about tolerance.  However, people tend to get (and I have no idea why this
is) passionate about either allowing fragment identifiers or NOT allowing
fragment identifiers.  In this way it becomes difficult or impossible to
form bi directional friendships (e.g. via a @rel attribute).  At TPAC this
year I heard one person arguing that hash URIs should always be used and
another that hash URIs should never be used.

Although the FOAF cloud is historically quite large, my experience of
current projects is that relatively few understand the value of using
fragment ids, even to the extent that they become not supported.  In the
latest WebID Identity spec authored by Tim, Henry and Andrei there was an
effort to address by having the examples show how fragment ids can be used

https://dvcs.w3.org/hg/WebID/raw-file/d21603d3972a/spec/identity-respec.html

So, ideally what I'd love to see is a document that I can point people to,
where if they were not too familiar with the motivation behind frag ids
they could see some examples, and choose one or the other, but support
both.  I hope that sheds a bit more light!

[1] http://indiewebcamp.com/Main_Page
[2] https://tent.io/about



>
> Jeni
> --
> Jeni Tennison
> http://www.jenitennison.com/
>
>
>
>
>