Re: [Turtle] Existing data , existing code.

On Wed, Mar 2, 2011 at 1:10 PM, Andy Seaborne <andy.seaborne@epimorphics.com
> wrote:

>
>
> On 02/03/11 15:46, David Wood wrote:
>
>> +1 to nice little features, and I'm not scared of (minor!) changes that
>> break BC because there aren't that many Turtle parsers out there.
>>
>
> There has been a lot of talk about maintaining already-deployed data. But
> there is also the issue of already-deployed code.  When there was last a RDF
> WG, the deployed base was much smaller, and a substantial part being run by
> active developers in semweb technology.
>
> Nowadays, I contend, there are more existing deployments and they do not
> run on the bleeding edge.
>
> We see question about versions of software several iterations in the past,
> and there are good reasons why people do not want to upgrade, or not
> upgrade, frequently.  If it works - why change it?
>
> Adding features to Turtle, or anything else, affects the existing software.
>  Whether there many Turtle parsers or not, there are a significant number of
> instances of them.
>

I agree with this sentiment.  The Turtle team submission obviously hit a
sweet spot among developers (more expressive, less verbose than N-Triples
but not as complex as N3) and has been widely enough implemented that it has
become a de facto standard.  I think our primary goal should be to
standardize that submission as version 1.0 of the language.  Any new
features, whether they are syntactic sugar or to support graphs/quads,
should be considered as either a new version of the language or a different
language that is a superset of Turtle.

-Alex