W3C home > Mailing lists > Public > public-html@w3.org > September 2007

Re: Spec review: "URI (or IRI)"

From: Robert Burns <rob@robburns.com>
Date: Sun, 2 Sep 2007 13:04:23 -0500
Message-Id: <EE594439-C783-4E71-8C11-05ADBD487185@robburns.com>
Cc: "public-html@w3.org" <public-html@w3.org>
To: Julian Reschke <julian.reschke@gmx.de>

HI Julian,

On Sep 2, 2007, at 12:36 PM, Julian Reschke wrote:

>
> Hi,
>
> I was reading through parts of Section 3, and noticed that the spec  
> seems to inherit outdated terminology from HTML4, and makes it even  
> worse by throwing in IRIs.
>
> For instance, in 3.7.4 (<http://www.w3.org/html/wg/html5/#the- 
> link>), it says:
>
> "The destination of the link is given by the href attribute, which  
> must be present and must contain a URI (or IRI)."
>
> First of all, the destination can be a relative reference as well,  
> so it should either say "URI or relative reference", or "URI- 
> Reference" (RFC3986, Section 4.1).
>
> Looking at <http://www.w3.org/html/wg/html5/#terminology>, it says:
>
> "For readability, the term URI is used to refer to both ASCII URIs  
> and Unicode IRIs, as those terms are defined by RFC 3986 and RFC  
> 3987 respectively, and as modified by RFC 2732. On the rare  
> occasions where IRIs are not allowed but ASCII URIs are, this is  
> called out explicitly. [RFC3986] [RFC3987] [RFC2732]"
>
> I find that confusing, in particular if other sections of the text  
> continue to say "URI (or IRI)". (Of course the reference to RFC2732  
> doesn't make any sense at all, as it has been incorporated into  
> RFC3986.)
>
> So, to include both URIs, IRIs, and URI references and IRI  
> references, the spec should really talk about "IRI  
> references" (including all of them), and point to the definition in  
> RFC3987, Section 2.2. If there's a need to use a different term for  
> that, I'd recommend *not* to use "URI", because that's really a  
> true subset of the things that the spec means to allow.

I wonder if we shouldn't be more specific and use the locator term  
when it applies. I'm not sure if any RFC deals with this, but many of  
the attributes that take IRIs in HTML5 really only take locators and  
not the broader identifier (is there an RFC on IRLs?).  If an  
attribute is going to take IRI as its data type, then I think we  
should clearly define what it means when the IRI is not a locator.  
For example, the proposal I made on associating attributions,  
citations, and references, where I defined how URNs (or would it be  
IRNs?) would be used[1].

So I think there are two axes to deal with: 1) ASCII  v Unicode and  
2) Locator and Name versus Identifier.

Take care,
Rob

[1]: <http://esw.w3.org/topic/HTML/AttrtibuCitaQuotationReferencing>  
(see also
  <http://lists.w3.org/Archives/Public/public-html/2007Aug/0076.html>)
Received on Sunday, 2 September 2007 18:04:44 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:49 UTC