- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Mon, 15 Sep 2003 12:01:43 +0100
- To: Charles White <Charles.White@networkinference.com>
- Cc: Ian Horrocks <horrocks@cs.man.ac.uk>, Jeremy Carroll <jjc@hpl.hp.com>, Mehrdad Omidvari <Mehrdad.Omidvari@networkinference.com>, Jack Berkowitz <Jack.Berkowitz@networkinference.com>, www-webont-wg@w3.org
Charles White wrote:
> Jeremy,
>
> Down below, you indicate that the relative URIs should not begin with / or .
> The cases I have found (10 or so different files) all have relative URIs that start with /
> See the example contained in my note below.
>
> So my question remains, is this legal syntax?
>
> chas
The syntax is legal, (as defined in RFC 2396).
I was suggesting that we could decide not to use this syntax in the OWL
Test Cases (it is somewhat obscure).
There are masses (many more than 10) of these in the section on harder lite
tests.
It was unintended on my part (I automatically generated these tests from
Sean's tests, and had recently updated the output module in Jena to use
relative URIs where possible:
http://jena.sourceforge.net/IO/iohowto.html#output
)
I believe that the particular relative URIs for owl:XXX and xsd:YYY are:
a) unnecessary
b) potentially confusing in an unhelpful way
(they depend on a http://www.w3.org/ base URI).
c) However, most are in approved tests, which most implementors do not have
difficulty with.
Personally I find a) and b) more compelling than c), but don't quite have
enough discretion to make the change; I will propose this as a change to be
discussed at a telecon.
Jeremy
Received on Monday, 15 September 2003 07:11:57 UTC