Re: SVG12: IRI Processing rules and xlink:href

* Andrew Sledd wrote:
>I have a question about interpretation and reference resolution, in
>particular about the reference xlink:href="". What does the reference
>resolve to? Is it valid for image in the SVG Tiny Profile? Is it valid
>for use and animation in the SVG Tiny Profile?

Hi Andrew, I trust the SVG Working Group will formally address all your
comments on SVG 1.2 and update the draft in response to them, but as the
answers from Mark and Jon on this issue weren't really correct, I guess
someone should clarify the situation.

The most important thing here that both Mark and Jon missed is that the
draft refers to IRI References as defined in RFC 3987 for the xlink:href
attribute, not to "URIs" or "relative URIs" as defined (or not) in RFC
2396. (XLink 1.0, which defines xlink:href, refers to RFC 2396 though,
SVG Tiny 1.2 is not compatible with XLink 1.0 or SVG 1.1 in this regard)

Under RFC 2396 the empty string would be neither a "URI" (RFC 2396 does
not actually define that term) nor a "relativeURI" (which, as Jon points
out, must be non-empty) or a "absoluteURI" (which must be non-empty
aswell). The correct term would be "URI-Reference" which is defined as

  URI-reference = [ absoluteURI | relativeURI ] [ "#" fragment ]

RFC 3986, which obsoletes RFC 2396, changes various things here. First,
URI-References are now defined as

  URI-reference = URI / relative-ref

the term URI (now defined) is an absoluteURI with an optional fragment
identifier (not allowed for absoluteURI in RFC 2396) and "relative URIs"
do not exist anymore, they are now called "relative references" to avoid
confusion (people assume that a "relative URI" would be a special kind
of "URI" but they are not). A Relative reference may also have the
optional fragment identifier (not allowed for relativeURI in RFC 2396).

RFC3987 refers to such references as "Relative IRI References". Relative
IRI References are resolved to IRIs as defined in RFC 3986 for relative
references. The empty string would basically resolve to the absolute-IRI
part of the "base IRI". So your interpretation

>I interpret the specs (SVG 1.2/IRI/URI) to say as follows:
>1) xlink:href="" resolves to a reference to the document itself (if no
>other xml:base attribute is explicitly given in the ancestors to the
>referencing element).

is not really correct either. Consider an example like

  <!-- document at http://example.org/svg/svg -->
  <g xml:base="http://example.org/base">
    <image xlink:href="" ... />
  </g>

In this case the xlink:href would resolve to <http://example.org/base>,
so e.g. getTraitNS('http://www.w3.org/1999/xlink', 'href') will return
an IRI canonically equivalent to that. This means the fragment above is
basically equivalent to

  <!-- document at http://example.org/svg/svg -->
  <g xml:base="http://example.org/base">
    <image xlink:href="http://example.org/base" ... />
  </g>

Part of your conclusion is correct though,

  When a URI reference refers to a URI that is, aside from its
  fragment component (if any), identical to the base URI (Section
  5.1), that reference is called a "same-document" reference.

same-document references should not cause new retrievel action, so
the IRI would dereference to the document that includes the xlink:
href attribute. Whether a reference is a same-document reference is
not affected by xml:base or its lexical representation though.

>2) In terms of IRI processing (Section 14.1.1) this is both a local
>reference and for the scope of <image> and <use> is an indirect circular
>reference.

I'm afraid there are too many errors in the draft to tell what the
Working Group might have had in mind here.
-- 
Björn Höhrmann · mailto:bjoern@hoehrmann.de · http://bjoern.hoehrmann.de
Weinh. Str. 22 · Telefon: +49(0)621/4309674 · http://www.bjoernsworld.de
68309 Mannheim · PGP Pub. KeyID: 0xA4357E78 · http://www.websitedev.de/ 

Received on Wednesday, 8 June 2005 19:25:40 UTC