[HarfBuzz] Why does kerning modify glyph advance?

[Tom Hacohen]

On Mon, 2011-02-14 at 14:42 +0000, Jonathan Kew wrote:
> Modifying x_offset alone would adjust the position of a glyph but not the position of any following glyphs, which is not the effect you want kerning to have. (OTOH, it's usually appropriate for things like accent positioning.)
Oh, good answer, thanks.
> (How would you expect to make use of caching the "x_advance of a char"? Did you perhaps mean glyph here, rather than character? *Character* advances are context-dependent even without the presence of kerning, as there might be contextual substitutions that give you entirely different glyphs.)
I'm aware of that situation, I was actually referring to caching of it
per strings so I'll be able to know the sizes of parts of the string
without needing to recalculate them, but nvm.

Thanks a lot,
Tom.