Patrick wrote:
> what I'm getting at is that I'd rather NOT see an exemption in the normative SC language at all, and a recommendation in the techniques for authors to switch to using inline SVG or similar, more robust techniques (as icon fonts are arguably a clever but non-semantic hack)
I think we agree, I just meant that with an SC in place there would be something that people then fail (when aiming to conform with 2.1), and several method of passing (where using SVG is probably better, but adding a semantic attribute is a reasonable fall-back).
Font-icons are not exempted as such, the issue is a side effect of users doing what they can to make text readable.
> But isn't the catch-22 still there, just codified?
I meant that there is no hook for user-agent font-changers, because there is nothing that web-sites fail they don’t put in place the hooks for user-agent scripts to use, so the user-agent scripts can’t avoid killing font-icons which hampers their usefulness, which prevents the whole thing being solved.
I believe the same applies to MathML, except that using particular mark-up then means it can easily be excluded from user-side scripts.
Cheers,
-Alastair
