W3C home > Mailing lists > Public > www-rdf-interest@w3.org > March 2002

Re: n3 wishlist: automatic namespace selection

From: Tim Berners-Lee <timbl@w3.org>
Date: Thu, 7 Mar 2002 16:43:14 -0500
Message-ID: <003301c1c621$15be54a0$84001d12@w3.org>
To: "Sandro Hawke" <sandro@w3.org>
Cc: <www-rdf-interest@w3.org>
I am happy to make a the space of prefixless words available
to the defualt namespace.

However, I am not happy to make soemthing in which there is
a search path for a matching  namespace.  This is unbeliveably 
prone to error in  large systems.  I could give you anecdotes if
you like - someone who spent weeks searching for a bug
just because ERRNO had been added to a FORTAN common
block by someone else, and took precedence in the scope rules.
It means that I can  introduce a "text" in a namesapce,
not changing anything else, and republiush the namespace,
and suddenly working systems break as they start to chose the
new tim:text instead of sandro:text.

(Talk to DanC about the evils of import * from xxxx in python)

I like the fact that with namespaces you definitively say
exectly which one you mean.  It is a pain to write them,
I know.

How about a sort of "import from"?

@use  text, pen, author, color <//http://address/of/one/ontology#>.
@use cost, total <//http://address/of/one/ontology2#>.
@keyword  this, forall, forsome, a.
@prefix defualt   http://www.w3.org/2000/10/swap/log.

where the "use" and "keyword"  cut out specific words,
leaving the rest to be from the default space.

Tim

----- Original Message ----- 
From: "Sandro Hawke" <sandro@w3.org>
To: <timbl@w3.org>
Cc: <www-rdf-interest@w3.org>
Sent: Thursday, March 07, 2002 12:13 PM
Subject: n3 wishlist: automatic namespace selection


> 
> I'd love to have cwm (etc) figure out which namespace (prefix) to use,
> from a list.   Something like
> 
>    @prefix foo: <http://normal/prefix#>
>    @import ont1: <http://address/of/one/ontology/>
>    @import ont2: <http://address/of/another/ontology>
> 
>    foo:a text "Hello".
> 
> and have it fetch the two ontologies, see if either defines a "text",
> and use it.  This is a pretty natural thing to do if you're already
> doing validation (which I think should be the default mode anyway).
> 
> There are several approaches; what do you think of the basic idea?
> 
> Aside from the convenience of not having to keep track of prefixes
> (when you don't feel like it), I think it would help motivate people
> to actually publish their ontologies.
> 
> The n3 keyword "a" could also then properly be a vocabulary term, from
> an implicitely imported n3-defaults namespace.


Actually "a" isn't just a shorthand for rdf:type, as it can only be used as
a predicate. But that could change. But other keywords like @prefix
and @forsome (coming) are special syntax.
 
>     -- sandro
> 
Received on Thursday, 7 March 2002 16:44:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:53 GMT