W3C home > Mailing lists > Public > public-cwm-talk@w3.org > July to September 2004

Re: Notation for sets in n3

From: Tim Berners-Lee <timbl@w3.org>
Date: Mon, 2 Aug 2004 18:05:56 -0400
Message-Id: <212C952E-E4D0-11D8-B1F7-000A9580D8C0@w3.org>
Cc: public-cwm-talk@w3.org
To: Yosi Scharf <syosi@mit.edu>

How about adding at least one comma to a list syntax to give set 
syntax, so that the same punctuation is used for lists.  The comma 
could be thought of as adding unorderedness.  (The other use of a comma 
occurs in RDF statements with multiple objects - and the objects are of 
course nor ordered)

  (,)  The null set
(  :a  ,)  The set with  only :a in it
( :a, :b, :c)  The set with :a , :b and :c in it

- Advantage is that lists and sets are similar and use up the same 
delimiters.

You might think that the "," should add order (as in a python triple or 
sequence) but in fact in most grammars order is normally important -- 
you can't just reorder the tokens -- so a special character to indicate 
unorderedness makes more sense, one could argue.

Tim

On Aug 2, 2004, at 15:06, Yosi Scharf wrote:

> Recently, it was decided that Cwm should have support for sets.
>
> This leads to the question, what delimeter should a set have in n3? 
> The problem is n3 already use every ascii delimeter I can think of.
>    (...) is taken for lists
>    {...} is taken for formulae
>    [...] is taken for anonymous nodes
>    <...> is taken for resources
>
> My father suggested ..., but being as that cannot be typed on a 
> standard keyboard, is unlikely to be useful.
>
> My personal opinion is something like {|...|} would be best. Something 
> like $(...) might also work. It seems no matter what we do, 
> readability is compromised.
>
> Yosi
Received on Monday, 2 August 2004 18:06:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:11:01 GMT