Re: Error handling for SVG and MathML in HTML

Erik Dahlström wrote on 03/19/2008 11:26:00 AM:
>
> On Wed, 19 Mar 2008 13:06:52 +0100, Sam Ruby <rubys@us.ibm.com> wrote:
>
> >
> > Erik Dahlström wrote on 03/19/2008 05:08:25 AM:
> ...
> >> I agree that nothing should prevent you from using correct SVG
fragments
> >
> >> inline in html.
> >
> > Agree in principle, but "nothing" is too strong.  Requiring HTML to
> > support
> > "<!ENTITY>" may not be necessary or appropriate.
>
> Agreed. I'll try to clarify what I meant with "correct SVG fragments".
>
> I find it an acceptable trade-off that not every single SVG ever created

> can be copy-pasted into an HTML document. However, any valid SVG document

> fragment that doesn't depend on having XML PI:s or DOCTYPEs prior to the

> root svg element to be a fully XML-wellformed SVG document, should be
> copy-pastable into HTML and should work IMHO. Or in other words: if you
> can copy-paste the svg root (and its children) into a new empty document

> and that is still a valid SVG document, then it should also be work when

> pasted into an HTML document.
>
> This is similar to how SVG inline in any other XML markup would work
> anyway, that is: you can't put XML PI:s in the middle of the document. I

> think it's a non-goal to allow that. If you need that then you should use

> an external file, plain and simple. The same argument holds for DOCTYPE
as
> well.
>
> So, given these constraints, is "nothing" still too strong? :)

Examples of a few things worthy of further discussion:

a) Psychotic use of namespaces[1]
b) <![CDATA[ ]]> (I'd say nice to have)
c) Attribute Value Normalization[2]
d) xml:space

- Sam Ruby

[1] http://lists.xml.org/archives/xml-dev/200204/msg00170.html
[2] http://www.w3.org/TR/REC-xml/#AVNormalize