Re: MathML for CSS profile

On 08/06/2011 21:58, William F Hammond wrote:

> For example, I find that something as simple as (x+y)^2 seems not to
> be handled well in browsers known for strong support of CSS.  See
> the screenshots in the HTML document
>
> http://www.albany.edu/~hammond/demos/forcss/overview.html
>

Thanks for the comments.
Your page has two examples, actually the second one doesn't conform to
the profile, it needs another mrow between the msup and the mfenced.

If that is added, it works as intended in Opera and IE (I tested it with
IE 10 but I think 9 should work as well).

It's unfortunate that it doesn't work in the current chrome, I tried it
with Chrome 14.0.803.0 dev-m, and got a result similar to your
screenshot. It appears it doesn't support all the css properties used
(in particular to flip the order of rendering inline table cells)

It's essentially an arbitrary and impossible choice, you can always
simplify the the css and get a wider implementation range but with
potentially worse rending.

The main normative force of the profile is to specify a subset of mathml
more amenable to css rendering. That applies whatever variant of css you
use.

Practically speaking, it appears that the best way to get good cross
browser support at present is to allow yourself some javascript
such as mathjax or the much simpler html5mathml, which just restricts
itself to the minimum required to enable mathml in the relevant browsers
(mathjax does this but also implements a mathml renderer in
javascript/css for those browsers without native mathml support, and
also offers Tex-like as well as MathML input syntax parsing).

David

Received on Thursday, 7 July 2011 21:59:36 UTC