Re: Implementation Report for Ruby LD Patch implementation

I can not remember if we discussed this point precisely.
On the one hand, I find it more elegant to have exactly one way to do it...
but on the other hand, being able to copy-paste a piece of SPARQL (or
SPARQL-like Turtle) would be nice.

On Wed, Aug 26, 2015 at 12:54 AM, Gregg Kellogg <gregg@greggkellogg.net>
wrote:

> Another piece of feedback on the grammar: The spec uses the @prefix
> notation from Turtle, but does not also allow the PREFIX form, inherited
> from SPARQL. This was added to make copy and paste between Turtle and
> SPARQL documents simpler, and the same rational might apply for LD Patch.
> Suggested grammar change would be:
>
> [2]     prologue        ::=     (prefixID | sparqlPrefix)*
> [6s]    sparqlPrefix    ::=     "PREFIX" PNAME_NS IRIREF
>
> My own grammar for Turtle uses the following:
>
> [4] prefixID    ::= PREFIX PNAME_NS IRIREF "."?
> [28t] PREFIX    ::= "@"?[Pp][Rr][Ee][Ff][Ii][Xx]
>
> which is more permissive, but reduces complexity.
>
> Of course, not including it might be intentional, but I’d consider adding
> it for simpler compatibility with both SPARQL and Turtle in the future.
>
> Gregg
>
> > On Aug 23, 2015, at 10:35 PM, Gregg Kellogg <gregg@greggkellogg.net>
> wrote:
> >
> > I’ve completed an implementation of a Linked Data Patch Format processor
> in Ruby, compatible with the RDF.rb suite.
> >
> > github: https://github.com/ruby-rdf/ld-patch
> > gem: https://rubygems.org/gems/ld-patch
> > earl report:
> https://raw.githubusercontent.com/ruby-rdf/ld-patch/develop/etc/earl.ttl
> > earl report (as HTML+RDFa):
> https://htmlpreview.github.com/?https://github.com/ruby-rdf/ld-patch/blob/develop/etc/earl.html
> >
> > As noted in the README on GitHub, this implementation uses a small
> variation on the EBNF Grammar:
> >
> > * The INDEX terminal was removed in favor of using INTEGER. As both
> INDEX and INTEGER can match the same terminals, it lead to an ambiguity
> that my LL(1) processor could not handle properly.
> > * The graph production is changed as follows:
> >
> > [18] graph ::= triples ('.' triples?)*
> >
> > This has the advantage of being context free, although does allow for
> repeated ‘.’ at the end of a graph, which certain tests ported from Turtle
> check for.
> >
> > The original production:
> >
> > [18]  graph   ::=     triples ( '.' triples )* '.’?
> >
> > is not context free, as ‘.’ is in the FIRST of two different
> productions; I think it should be replaced in the version I used for the
> grammar to be context free.
> >
> > This is a stand-alone processor, and needs to be integrated with an LDP
> server to be useful, such as https://github.com/ruby-rdf/rdf-ldp.
> >
> > Gregg Kellogg
> > gregg@greggkellogg.net
> >
>
>