Re: SVG and MathML in text/html

On Mar 15, 2008, at 17:16, Ben Boyle wrote:

> Scenario:
>
> I am using text/html. You've sold me on not deploying web content as  
> xhtml.
> I've embedded that html snippet, and I've made that error in the code.
> Unlike me to forget quotes of all things, but there you go.
> I'm seeing the red circle stroked blue, it's all cool.
>
> Suddenly I am enlightened that I should use an SVG file and link to
> it, rather than embed and serve this same snippet of SVG on every
> page. I dutifully move the SVG code into logo.svg, link it with an img
> tag and... what the? The circle went black!

Nope, you get the YSoD.

> These things are authoring nightmares. Just don't do it. Consistency  
> please!

You are basically asking the well-formedness "nightmare" to happen in  
*every* case instead of it happening only *if* a certain reuse  
scenario actualizes. You are effectively asking for by-design  
brittleness *just in case*, so the brittleness is there even when the  
reuse scenario *doesn't* actualize.

I much prefer the alternative of doing just-in-time XML conversion if  
the need to actually have XML actualizes. That way, those for whom the  
scenario is irrelevant aren't bothered for nothing.

>> How is that worth the complexity of having two different attribute
>> tokenization rules that get toggled in mid-parse?
>
> It's funny (ironic) you say that, because from my point of view you
> are proposing the complexity of two different attribute tokenization
> rules that get toggled depending on the source serialisation.

Well, you cannot make the HTML kind of attribute tokenization go away.

> We must be from different worlds ~:)


I live in the world of considering what kind of runnable code the  
opinions expressed here would translate into.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Saturday, 15 March 2008 15:43:54 UTC