Re: [xml-dev] SVG and MathML in HTML5

[David Carlisle <davidc@nag.co.uk>]

On 28/05/2011 09:58, Jesper Tverskov wrote:
> As we can read in the HTML5 spec, we can nest SVG and MathML inside
> not well-formed HTML5 served with mimetype "text/html". We can also
> read that SVG and MathML don't need to be well-formed:
>
> http://dev.w3.org/html5/spec/Overview.html#mathml
> http://dev.w3.org/html5/spec/Overview.html#svg-0
>
> The spec asks browsers, but it only asks, to provide a way to export
> not well-formed SVG and MathML in a well-formed way in order not to
> cause too much trouble for XML tools!
>
> As I see it, we will soon have a situation where most of the SVG and
> MathML used in the world, and that actually works in the browsers, are
> no longer XML.
>
> HTML5 also allows for HTML and SVG and MathML to be in their proper
> namespaces implicitly. You don't need to declare them!
>
> Personally I don't like that a new W3C spec is allowed to undermine
> several of the most fundamental concepts of XML. It might make things
> marginally easier in HTML but it is bound to cause a lot of trouble in
> the XML world.
>
> At least it will be more difficult to teach people what an XML
> application is. Some of them also work in HTML and don't have to be
> well-formed.
>
> It will be even more difficult than today to make sense out of
> namespaces. You must always declare them and when you want to access
> markup using namespace with programming, you must also always declare
> them in that programming language. Except that in HTML you don't need
> to declare them, etc., etc.
>
> Also it is not funny that most SVG and MathML in the future probably
> no longer can be edited and manipulated in XML Editors unless the
> markup is first made well-formed, namespaces added, etc.
>
> Why have we allowed all this to happen?
>
> Cheers
> Jesper Tverskov
> http://www.xmlplease.com
>

Actually, I don't think that not requiring well formed input is too bad, 
in fact it's almost certainly the right thing to do in an html context.

What is potentially harmful is that as specified (for no reason that 
makes any sense to me) is that well formed mathml/svg with well formed 
html content parses completely differently.


http://www.w3.org/Bugs/Public/show_bug.cgi?id=9887

However we managed to restrict this to some cases that can be more 
easily avoided... so we allowed that bug to close.

On balance though, having mathml in html is a massively good thing.
Mathml has had good takeup in xml toolsets mostly any tools requiring 
math markup are converging on mathml, but using it "in the wild" has 
always been hampered by the fact that it was tied to using xhtml (xhtml 
as specified, served as xml, not the "appendix C monstrosity of xhtml 
served as text/html) being able to just drop the mathematics into a 
normal web page (so use in blogging and other cms systems for example)

I posted a blog article on svg and mathml in html  last year:

http://thenumericalalgorithmsgroup.blogspot.com/2010/09/html5-possible-implications-for.html


David