Re: [css3-text] grapheme clusters across element boundary

I've been looking into an related instance of this problem (in webkit)
related to arabic shaping across an element boundary [1]. There is clearly
an authoring requirement to be able to style individual characters that
participate in a grapheme cluster or that contribute shaping context.

[1] https://bugs.webkit.org/show_bug.cgi?id=6148

On Sat, Jan 14, 2012 at 10:53 AM, Kang-Hao (Kenny) Lu <
kennyluck@csail.mit.edu> wrote:

> There seems to be endless trouble if we consider corner cases involving
> grapheme clusters across element boundary, i.e., for example, the
> following case is not, and might forever not be, implemented interoperably:
>
> data:text/html,t<u>&%23x308;</u><s>est</s> (where &#308; is U+0308
> COMBINING DIAERESIS)
>
> I suggest we explicitly mark this kind of situation as undefined in the
> spec near the definition of a grapheme cluster, as I believe this is
> better than saying nothing at all with regard to this.
>
> Perhaps CSS Text beyond level 3 (or CSS Fonts?) should consider
> addressing use case like coloring the combining mark separately form the
> base character[1] although I wonder whether standardizing this behavior
> might hard competition in the space of font systems.
>
> [1] https://www.w3.org/Bugs/Public/show_bug.cgi?id=13502#c11
>
> Cheers,
> Kenny
>
>