Re: Escape sequences (SPARQL and Turtle)

RDF-WG is still discussing escape sequences.  One of the areas of debate 
is being able to put additional characters into prefixed names, 
motivated by (1) Facebook Open Graph Protocol, which naturally lead to 
":" in the local part of prefix names and (2) convenience working with 
generated IRIs, e.g. R2RML.

OGP:
<html xmlns:og="http://ogp.me/ns#">
....
     <meta property="og:audio:title" content="Amazing Song" />
....

R2RML IRI template:
http://data.example.com/employee={EMPNO}/department={DEPTNO}


An original proposal to use unicode escapes (\u....) for this has 
resulted in some pushback.

A recent character-escape approach is also in this space [1].

PREFIX og:  <http://ogp.me/ns#>
...
   og:audio\:title
...

== Progressing SPARQL 1.1

I don't think that "at risk" in CR works very well for "A or B" choices 
because whichever we choose, some implementation might have to change. 
If it's "A or not A" feature, then anything implementing A isn't made 
non-conformant, it's merely has an extension.

A way forward??

1/ SPARQL's \u processing remains the same.  This means we are either 
adding something or not  We are not changing something - so closer to 
the charter.

2/ Decide close to 2LC whether to put in an at-risk feature for \: etc. 
into 2LC.  Do this as close to 2LC as possible, seeing if RDF-WG has 
made progress.

If it gets through 2LC, we keep this feature in CR still "at-risk". 
Again, at CR, review whether RDF-WG is settled in this area.


We can put some advice in the SPARQL/RDF-1.1 page if the feature is 
pulled.  As this advice page has no status, it won't get much prominence.

 Andy

PS I have prototyped [1] in SPARQL and nothing broke nor were any tests 
affected.

[1]
http://lists.w3.org/Archives/Public/public-rdf-wg/2011Nov/0234.html

Received on Sunday, 27 November 2011 21:19:16 UTC