Re: [Turtle] Misc initial thoughts

On 03/03/11 07:21, Steve Harris wrote:
> On 2011-03-02, at 17:31, Nathan wrote:
>
>> Richard Cyganiak wrote:
>>> Andy, On 28 Feb 2011, at 20:36, Andy Seaborne wrote:
>>>> A data-format N-triples / N-Quads would be a subset of Turtle,
>>>> with the same IRI resolution rules and same syntax for IRI
>>>> tokens.  And in UTF-8.
>>>>
>>>> As these formats are used as dump formats, pinning down
>>>> details would be a help to data publishers and consumers.
>>>>
>>>> A MIME type which is not text/plain would be helpful.
>>> I think having a proper spec for this “N-Triples done right” is
>>> a great plan, including support for quads, IRI resolution,
>>> UTF-8, and proper media type. However I wouldn't necessarily see
>>> it as a subset of Turtle. I'd prefer for Turtle to remain as it
>>> is defined now, as a triples-only format without
>>> multigraphs/quads.
>>
>> quads != triples surely, perhaps there needs to be two then,
>> N-Triples and N-Quads.
>
> Agreed. I don't like the idea of having formats in the wild which
> might reasonably be expected to contain triples, or quads and there
> being no easy way to tell before you start parsing it. Additionally,
> the ease of parsing is the main selling point of N-T/Q, and no
> knowing if there will be 3 or 4 columns makes it trickier.

+1 here as well.

POSTing to a graph can only be triples.

GETting triples, you know you can put them in a graph in a dataset.  It
is unclear what to do if you start getting quads part way through.

N-Quads serializes a datasets, default graph is no 4th field.
N-triples serializes a graph.

> FWIW, N-Quads seems quite popular for data.gov.uk data, and
> N-Triples/N-Quads are good lowest common denominator formats, for
> bulk dumps and the like.

I can imagine N-triples only parsing to get a small footprint.  Turtle
isn't technically complicated but neither is it trivial to parse.

> I would like to see an update to make UTF-8 legal though, and some
> clarity on BASE URI resolution.
>
> - Steve

	Andy

Received on Thursday, 3 March 2011 10:31:01 UTC