Re: <title> containing markup

Doug Schepers:
> No, it doesn't impact it.  I knew at the time that the content of both  
> <title> and <desc> could contain markup in SVG 1.1, but that it was  
> underspecified.  After looking at the existing capabilities of existing  
> SVGT1.2 UAs, only a minority of them handled markup other than SVG, so I  
> judged that *for Tiny* (which is more targeted at devices with limited  
> resources) it made sense to restrict the content of <title> to text,  
> which would give us a clear way of handling the content.  I always  
> planned that for SVG 2.0, we would allow markup, as we do in SVG 1.1,  
> but with a clearer model for handling it.

Acknowledged.

> I don't think we need to go down the absolutist path of HTML5, though,  
> where we restrict what markup can be used and explain exactly how to  
> parse it.  Most other languages are grown-ups, and can describe for  
> themselves how they need to be handled.  We could explicitly mention how  
> some particularly useful languages (X/HTML, DocBook, RDF, XSL) might be  
> dealt with, and make a general rule that only text-container elements  
> should be used (no images or iframes, no script excution, etc.), no  
> matter what the hosted language is.  For <desc>, at least, this should  
> also included traditional structured-text elements, like paragraphs and  
> lists.

So you are in favour of having SVG <title> in text/html be parsed as “in
foreign content”, which would allow phrasing content to be used?  Would
we make the inclusion of HTML phrasing elements in SVG <title>
conforming but any other elements non-conforming?  (Or make any SVG,
MathML or non-phrasing HTML element non-conforming?)

> > I’m wondering also if someone could tell me the exact i18n
> > problems allowing markup rather than plain text solves. Do the
> > Unicode bidi control characters (like RLE, PDF, etc.) not allow you
> > to do everything you need to? If there isn’t a compelling reason
> > here to allow markup, then I’m in favour of keeping our request
> > for it to be parsed as RCDATA in HTML 5.
>
> We should bring in an i18n person to advise us here.

I’ll mail an i18n list and ask.

> One factor in favor of allowing markup is having a single clear model  
> used for both <title> and <desc>.

Do you mean to have both <title> and <desc> allow the same child content
(in text/html and/or XML SVG)?  Or to have a single, coherent model
(such as having <title> allow only phrasing HTML elements while allowing
<desc> to have any HTML elements)?

The other point of consistency requested by some is that since HTML
<title> is parsed as RCDATA, SVG’s (in text/html) should be too.  IMO if
it can be shown that SVG <title> would benefit from having element
children, then I think this consistency with HTML <title> isn’t as
important.

-- 
Cameron McCormack ≝ http://mcc.id.au/

Received on Monday, 23 March 2009 05:45:08 UTC