Re: machine consumable SPARQL grammar

On Mon, Nov 21, 2005 at 09:51:00PM -0600, Dan Connolly wrote:
> 
> On Tue, 2005-10-18 at 09:44 -0400, Eric Prud'hommeaux wrote:
> > On Tue, Oct 18, 2005 at 01:20:38PM +0100, Seaborne, Andy wrote:
> > > 
> > > 
> > > 
> > > Dan Connolly wrote:
> > > >So our grammar is now LL(1).
> > > >http://www.w3.org/2001/sw/DataAccess/rq23/#grammar
> > > >
> > > >But to directly consume the current draft (1.511 ) by machine,
> > > >I guess you have to copy and paste the table or something.
> > > >I think we have yacc and n-triples versions of the grammar.
> > > >Please add those to the draft, OK, EricP or Andy?
> > > 
> > > No problem.  It isn't just yacc anyway.
> > > 
> > > (N-triples?)
> > 
> > The N-triples-serializing code is probably a little fettid now.
> > 
> > > I'd also like to see that the generated parsers can in fact parse the test 
> > > suite syntax tests as a sanity check.
> > 
> > I have lexer probs with the python one (don't have the esaping right,
> > DanC, could you massage the .g file and tell me what escaping diffs
> > are required?).
> 
> Er... "the .g file"? Whazzat? Pointer? Clues?

The yapps language spec:
  http://www.w3.org/2005/01/yacker/uploads/SPARQL/SPARQL.g?lang=perl

invoked with this Makefile:
PARSERS = SPARQL

parsers: $(PARSERS)

$(PARSERS): %: %.g
	    yapps $< SPARQL
	    sed "s/>>sys.stderr, 'Args:  <rule> \[<filename>\]'/parse('Query', stdin.read())/1"  <$@ >$@.t
	    mv $@.t $@

> >  I periodically run the SyntaxTests through the perl
> > and C versions. Can do this again. Is now the time?
> > 
> > > >
> > > >i.e. check them into the rq23/ directory and add a link
> > > >from the #grammar section.
> > > >
> > > >In Andy's repy to the comment, there's a pointer to yacker,
> > > >and I can follow my nose from there to a list of grammars
> > > >  http://www.w3.org/2005/01/yacker?action=list+grammars
> > > >but it's not clear which is the relevant one.
> > > 
> > > This is true - Eric is there someway we can:
> > > 
> > > 1/ Delete unused grammars
> > 
> > In order to not get distracted by yaccer just now, I'll do this from
> > the command line. Are there any other than rq23final that you want to
> > save?
> > 
> > > 2/ Ensure all the generated parsers match the grammar.  At the moment, each 
> > > needs to be explicitly remade, right?  Flagging when the gramamr is newer 
> > > than one of the parsers and a "make all parsers" option woudl help.
> > 
> > This would require a bit of re-eng. Right now, the differen version
> > overright each other (not a great design).
> > 
> > > The grammar is rq23final (no "-") although that is prior to the change made 
> > > last week s/isURI/isIRI/g which changed the grammar in rq23 as well as all 
> > > the text use of isURI.
> > > 
> > > >
> > > >Is this relevant to life as we now know it?
> > > >  http://www.w3.org/2001/sw/DataAccess/rq23/grammar.yy
> > > >  grammar.yy,v 1.5 2004/11/28 08:28:39
> > > >
> > > >
> > > 
> > > 	Andy
> > > 
> > 

-- 
-eric

office: +81.466.49.1170 W3C, Keio Research Institute at SFC,
                        Shonan Fujisawa Campus, Keio University,
                        5322 Endo, Fujisawa, Kanagawa 252-8520
                        JAPAN
        +1.617.258.5741 NE43-344, MIT, Cambridge, MA 02144 USA
cell:   +81.90.6533.3882

(eric@w3.org)
Feel free to forward this message to any list for any purpose other than
email address distribution.