(In reply to comment #19)
> (In reply to comment #18)
> > I confess I'm totally lost. Are you using PangoXft with cairo?!
> >
>
> No. The file is possibly misnamed now. The intention was to add a fast path
> that (at the very least) *avoids* using Pango for measuring the types of
> strings we need to measure. This second path was wedged in with the existing
> Pango code but it uses Xft alone to measure and get the glyph indices. Glyphs
> are then (still) rendered with the cairo font API.
I perfectly understand that. But it doesn't change a thing. You are using PangoXft (pango_xft_font_get_font for example) and cairo, bridging in between yourself, instead of using PangoCairo. I won't comment on the level of pain that brings, and let alone what a broken design it is, but it does mean that I can't and won't spend time making it faster as all my focus is in PangoCairo...
(In reply to comment #19)
> (In reply to comment #18)
> > I confess I'm totally lost. Are you using PangoXft with cairo?!
> >
>
> No. The file is possibly misnamed now. The intention was to add a fast path
> that (at the very least) *avoids* using Pango for measuring the types of
> strings we need to measure. This second path was wedged in with the existing
> Pango code but it uses Xft alone to measure and get the glyph indices. Glyphs
> are then (still) rendered with the cairo font API.
I perfectly understand that. But it doesn't change a thing. You are using PangoXft (pango_ xft_font_ get_font for example) and cairo, bridging in between yourself, instead of using PangoCairo. I won't comment on the level of pain that brings, and let alone what a broken design it is, but it does mean that I can't and won't spend time making it faster as all my focus is in PangoCairo...