W3C home > Mailing lists > Public > www-rdf-interest@w3.org > January 2001

Re: Dropping the redundant colon in N3

From: Tim Berners-Lee <timbl@w3.org>
Date: Tue, 23 Jan 2001 17:15:52 -0500
Message-ID: <05cf01c0858c$436fa600$0301a8c0@w3.org>
To: "Sean B. Palmer" <sean@mysterylights.com>, "Seth Russell" <seth@robustai.net>
Cc: "RDF-IG" <www-rdf-interest@w3.org>
This is a good point.  The colons  drive most right-minded people
crazy, unless one is mixing languages anyway in which case
the prefixes help.

The problem is this.  There is a clash with keywords.  There are of course
a limited number of keywords, such as bind, @prefix, a, has, of, to which
I am tempted to add with time (which?...).  Any language which uses
a set of special keywords in a space of identifiers open to the
user has a problem with evolution,  If I allowed unprefixed identifiers
now then any keyword I add later could invalidate old N3 documents.
One possibility, rather clumsy, was for unprefixed names to be unavailable
unless the document declares exactly which keywords it will use.

@keywords a, has, which, of .
x a rfd:Class. y a x.

I don't want to move on this until I am happy that the outcome will be safe.

Tim

----- Original Message -----
From: "Sean B. Palmer" <sean@mysterylights.com>
To: "Seth Russell" <seth@robustai.net>
Cc: "RDF-IG" <www-rdf-interest@w3.org>; "Tim Berners-Lee" <timbl@w3.org>
Sent: Saturday, January 20, 2001 4:43 PM
Subject: Re: Dropping the redundant colon in N3


> > i reread the primer quickly and I don't see anything about "keywords"
>
> Try: http://www.w3.org/DesignIssues/Notation3
> For example, you could have :x has :y of :z, which if you got rid of the
> colons would be x has y of z. Actually, I don't like that, because what if
> you had ":hasAuthor", then it doesn't make grammatical sense (e.g. "has
> hasAuthor of", or even worse "is :hasAuthorOf of"); but TimBL said that is
> optional (I suppose to make it a bit more legible to humans). I'd take 'em
> out if it were up to me, and let the properties do the talking :-)
>
> > But your right, if the intent is to allow any key words whatsoever
> > and just have the parser ignore them, and if that is more important
> > than worrying about the messy redundancy, then the colon would
> > be necessary.
>
> I think the colon is there due to the fact that it fits in neater with the
> namespace spec. I wouldn't make much sense to have:-
>
>      @prefix <URI>
>
> Because you would be expecting something to bind. Maybe if the ns alias
> (prefix) to bind were in quotes, ala.:-
>
>      @prefix "myprefix:" <URI>
>
> then you could have
>
>      @prefix "" <URI>
>
> But it gets a bit messy. I don't know really, I think it's alright the way
> it is. "Too many cooks spoil the broth", and all that :-)
>
> --
> Kindest Regards,
> Sean B. Palmer
> @prefix : <http://infomesh.net/2001/01/n3terms/#> .
> [ :name "Sean B. Palmer" ] has :homepage <http://infomesh.net/sbp/> .
>
Received on Tuesday, 23 January 2001 17:26:50 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:47 GMT