- From: Andy Seaborne <andy.seaborne@epimorphics.com>
- Date: Wed, 28 Mar 2012 18:13:50 +0100
- To: RDF-WG <public-rdf-wg@w3.org>
... and not in the blank node labels or in the prefix part.
Tokenizing ambiguities are different from grammar ambiguities - there
are no ambiguities as such; there are resolution rules. In a tokenizer
"longest match wins", and if the same length, it's common that the first
one wins.
Changing the SPARQL grammar did not break anything additional to the
issue of the empty string prefix name and no white space separating
adjacent uses of prefix names.
------------------
@prefix a: <http://example/NS/> .
@prefix : <http://example/NS/> .
a:b:c:d .
------------------
The datatype case
"1"^^xsd:integer:foo:bar
can't occur for other reasons (literals as object only, no following
prefix name possible).
As a bonus, I tried tweaking a Turtle parser and skimmed the
not-official test cases.
It looks like bad-07.ttl now fails before the point it was supposed to test.
[[
This is a technical investigation and not an endorsement.
]]
Andy
Received on Wednesday, 28 March 2012 17:14:25 UTC