Re: [CSS3-text] text-underline-position and superscript

Sorry for following up on myself again. I'll blame it being early
morning, but let me reorganize myself a bit and restate how the two
ideas are related:

==================================================

There are two generalized principles common to Charles' and my ideas,
and the two general principles are that:

1. We need some way to specify that, in some situations, no matter
what the glyph-specific underline position is, we want to keep a
constant underline position for some logical grouping of characters.

(1a) For superscript/subscripts: underlines don't move up/down due to
the super/subscripting

(1b) For Chinese: underlines don't move up/down when there are Latin
or other non-CJK characters in the sequence

2. As a corollary of the above, we need some way to specify that
underlines are always visually disjoint if they are semantically
marked up as separate.

(2a) For superscripts/subscripts: The logical markup is provided by
SUP or SUB and we make it clear that we want the underlines to move
up/down along with the super/subscript

(2b) For Chinese: The logical markup is provided by U and we make it
clear that the two adjacent underlines should never run into each
other

(2c) The Chinese use case could also potentially be useful for
non-Chinese situations

What Charles proposed are ways to specify how the constant underline
position in #1 should be determined, and to specify how a non-constant
underline position in #2 can be explicitly specified for superscripts
and subscripts. Perhaps there can be ways to get rid of the proposed
keywords, but his proposal is a good analysis (without considering the
requirements for the Chinese typography) of what we will need to deal
with when we need the browser to figure out a constant position for
the underlining.

Charles did not explicitly specify a use case for "pixel positioning",
but I suggested it as a possible fix for incorrect underline position
in Chinese. The above also shows that the counter-proposal of
correcting the underline positions in CJK fonts (which still should be
corrected, since this affects also word processors) alone will not be
a complete fix to the Chinese problem. Personally, I envision "pixel
positioning" to be usable as a workaround for both problem #1, and
problem #2 when we are dealing specifically with superscripts and
subscripts; it may not be a perfect solution but this could be what
Charles had in mind, *especially* if you don't want the proposed
additional keywords.

-- 
cheers,
-ambrose

does anyone know how to fix Snow Leopard? it broke input method
switching and is causing many typing mistakes and is very annoying