Re: add Turtle examples to specs

On 3/13/06, tim.glover@bt.com <tim.glover@bt.com> wrote:
>
>
> But syntax does matter. For example, I am told that Newton's formulation
> of calculus was impenetrable, and progress in physics owes a lot to the
> alternative syntax by Leibnitz. And compare doing arithmetic using Roman
> Numerals with positional notation.

To play the Devil's Advocate, I'd point out that the intended purpose
of the each notation differs, that suitability for the task needs to
be considered.

I'm sure you're right about the impenetrability of Newton's syntax (he
was an alchemist after all), but I'm not sure either syntax is much
good for the command line.

The Wikipedia tells me that Roman numerals "appear to derive from
notches on tally sticks, such as those used by Italian and Dalmatian
shepherds into the 19th century". I imagine the shepherds would have
had problems with positional notation, having to whittle the stick
afresh for every new sheep. Baa.

> Pragmatically, if RDF has a future it WILL be given a simple syntax - de
> facto and by practitioners, in the face of howls from the W3C if
> necessary!

I don't disagree, and the simpler syntax of Turtle does seem pretty
good for human-legibility. But for machine comms it's a different
matter. There are issues with RDF/XML in this context, notably the
impedance mismatch to most XML tools. But this is something which
would be the case to a greater or lesser extent whatever the syntax
looked like because graphs > trees. RDF/XML is fine between RDF
systems, and interfacing with the XML and HTML worlds is possible
(especially when SPARQL is available). Also RDF/XML isn't bad as a
read-only language for debugging purposes (compare with SOAP).

So, trying to speak pragmatically, my tendency would be to encourage
Turtle (especially around education/outreach) but accepting that
RDF/XML (or domain-specific XML where appropriate) will be there on
the wire. I suspect human resources would more effectively be used in
building stuff than trying to standardize on a different XML
serialization.
 (I'd place things like GRDDLable XML and Embedded RDF in a different
bag, time will tell what role they might play in RDFmachine-RDFmachine
comms).

Cheers,
Danny.

--

http://dannyayers.com

Received on Monday, 13 March 2006 13:17:11 UTC