Re: Turtle implementation report for RDF::Trine

On Thu, 2013-04-25 at 17:23 +0800, Gregory Williams wrote:
> Dave,
> 
> Another follow-up on this issue:
> 
> On Apr 17, 2013, at 10:22 PM, David Wood <david@3roundstones.com> wrote:
> 
> > There is a note in ISSUE-1 that shows that the SPARQL 1.1 syntax was "practically frozen" by the time the RDF working group was established (23 Feb 2011, 16:28:05).  Unfortunately for Turtle syntax, we had very little ability to change SPARQL.
> > 
> > There is also a note that the working group resolved that SPARQL and Turtle syntax should be "the same except for well-motivated (and small) exceptions." (resolved at 13 Oct 2011, 17:24:43 UTC)
> > 
> > Both of those notes suggest @prefix and PREFIX syntax alignment.
> 
> The early discussion of ISSUE-1 seems to be only about the triple(-pattern) syntaxes of Turtle and SPARQL. In fact, Richard Cyganiak brought this up explicitly by saying "The resolution wast that the *triple pattern syntax* should be the same … This shouldn't be understood to include BASE and PREFIX, I think" [1]. Which got agreement from at least Sandro[2]. At some point people started discussing PREFIX and BASE again, but it's not clear to me that any resolution of ISSUE-1 or its motivation that triple-pattern syntaxes be aligned should have bearing on this issue.
> 
> I certainly don't think that arguing that the WG resolution to align the triple pattern syntaxes should outweigh what is, in my opinion, a valid technical and usability argument against having two separate syntaxes for PREFIX and BASE. Especially when there wasn't exactly overwhelming support for adding this to the spec to begin with. Looking at the 16 May 2012 minutes[3], I see 11 of 17 abstaining (0) votes, 3 votes for (with an additional +0.2) and 2 votes against (-0.999 and -0.7). (I'm not going to try to parse the difference between people's using signed zeros.)

Exactly.  There are certainly things about the triple syntax I don't
like at all (path query idiocy), but aligning those is a reasonable
thing to do and I support it despite the ugliness.  That is "aligning":
making the two very similar things equivalent.

However adding new redundant directive syntax isn't; there is no
existing practice this breakage fixes, and it should be pretty obvious
that having redundant and inconsistent forms for the same thing in a
language is not so great.  There is no alignment because the old forms
aren't going anywhere, there is only cruft - cruft which implementations
clearly should not be writing in any case.

The W3C bureaucratic process is being confused with reality too much
here.  Turtle is an established language that has been around for many
years; the SPARQL WG happening to finish first (not giving a damn about
Turtle in the process) does not change that.

It seems there is nowhere near enough support for making such a change
to Turtle which, like it or not, has had an established single form for
directives for years.

-dr