Manuel Mall wrote:
What about character composition/decomposition?
Good question? Where is the answer?
Lets clarify the problem first. Let's say the input contains
the sequence U+0061 U+0308 (latin small a, combining diaresis),
the font has a glyph for U+00E4 but not U+0308. Obviously,
putting the precomposed character U+00E4 into the output is
a smart move. Where should this transformation occur: output
generation, renderer, layout stage? A slight problem is that
the width of U+00E4 may be different from U+0061.