[whatwg] Supporting MathML and SVG in text/html, and related topics

[Øistein E. Andersen]

On Thursday 10th April 2008, Ian Hickson wrote:

> SVG radicals aren't typographically acceptable either.
> You really want to use fonts for this.

Current browsers are clearly better at rendering TrueType
and PostScript fonts at small sizes than equivalent shapes
expressed as SVG paths.  (This may or may not or may only
in part be related to hinting as I never tested this using
hinted and unhinted versions of the same font, but I
suspect that hinting does not account for everything.)
Poor or even abysmal on-screen rendering made me abandon this
approach last time I considered maths-to-SVG conversion.

This particular problem would however be less of an issue with
bigger and/or more geometrical shapes, and I would consider
TeX's construction of, e.g., vincula (horizontal lines) by
overstriking of tiny bits from a font to be an artifact of
not being able to intermix text and graphics freely rather
than to result from intrinsic aesthetic superiority of the
`everything-from-fonts' approach.

Now that Safari for Mac supports custom fonts using
@font-face and other browsers will follow suit, using
fonts for text and operators and SVG graphics for big
delimiters and geometric symbols would seem to be a
reasonable approach, and I would be interested to know
what might make SVG radicals `typographically unacceptable'.
(Obviously, fonts and SVG elements must be coordinated.)

<http://coq.no/musica/it> illustrates the concept for
musical notation (SVG lines to draw the staves combined with
a font for the clefs and accidentals), and I think SVG
would also be appropriate for ties in musical
notation, bonds in chemical 2D formulae, &c. to 
achieve high-quality, typographically sound rendering.
Am I naïvely overlooking an inherent problem with SVG?

-- 
Øistein E. Andersen