Re: N-triples (1.4)

>>>Satoshi Nakamura said:
<snip/>

> If n-triples containing characters other than us-ascii are escaped, it
> would not be viewed with a normal viewer or an editor. It may be a problem
> for people who use a language other than English.
> 
> However, there are some problem.
> 
> 1. Is it possible to using encoding other than utf-8?
> 
> For example, some servers or gateways may make a content negotiation and
> convert its charset, if n-triples uses text/* content-type. What happen if
> the server convert its encoding to other than utf-8 and replace charset
> parameter?
> 
> 2. What happen if n-triples are transfered using 'Content-Type:
> text/plain'?
> 
> MIME specification says that if there is no charset parameter, it must be
> treated as us-ascii, and HTTP says iso-8859-1. Is it an error?
> 
> 
> To avoid these problem, RDF uses XML as a container, and it solves these
> problems. So, it's may be nonsense to discuss about transfering n-triples.
> In this meaning, escapeing all characters other than us-ascii, and
> handling n-triples as us-ascii text seems good idea.

I agree with your last comments - we use XML for doing this properly
and this is a simple text/plain format that *can* encode all
characters, and is indented primarily for developing and testing RDF
syntax reader/writer code, and as a concrete syntax for our formal model.

This is not intended for end users to type as a new RDF syntax.

[ At least not yet... if we do change our minds about this later, we
can revisit it later and use, for example, UTF-8 since it is a
superset of US-ASCII, the 0..127 code points are the same.  But I
don't want to start that discussion now. ]

So I am going to change the N-Triples document to be a US-ASCII
format (chars 0..127) and the \-escapes to allow encoding of the
higher code points.

Dave

Received on Tuesday, 24 July 2001 06:17:05 UTC