Re: Review of N-Triples draft

On 07/15/2013 04:27 PM, Gregory Williams wrote:
> [ . . . ] I was hoping for insight into why
> "Canonical N-Triples" shouldn't just say "one space following every
> term" which I believe to be simpler both in describing the
> constrained grammar and in implementation.

+1

I personally *hate* the fact that Turtle and SPARQL allow a period 
immediately after the object, thus leading to this abomination:

   { :foo :bar 1234. }

That last term is actually parsed as an *integer* when in any normal 
programming language it would have been parsed as a floating point number.

I would much rather have canonical n-triples discourage the habit of 
omitting a space before the period by *requiring* a space before the 
period (even though n-triples requires all literals in long form).

Requiring a single space after the object also makes it easier to split 
the line on whitespace and get all of the (subject, predicate, object) 
terms.


[ . . . ]

Re: media type:

> I've always found text/* to be much easier to deal with as 1) it's
> trivial to force a link to download in a browser with an extra
> key-press and 2) it *allows* peeking inside the file in the browser
> if desired (which is often impossible with application/*).

+1

A couple of other comments:

1. Section 6.1 says:
https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-turtle/n-triples.html#sec-parsing-terms
"The string matching the second argument after '_:', is a key in 
bnodeLabels."

Shouldn't that say something like "The string after '_:', is a key in 
bnodeLabels."?

2. Typo:  s/lexicalf orm/lexical form/

David

Received on Monday, 15 July 2013 21:30:33 UTC