[cairo] API Shakeup: Eliminating cairo_show_surface

[Bill Spitzak]

Carl Worth wrote:

>    Owen made the recommendation to add an (x,y) offset to
>    cairo_set_source_surface so that usage becomes:
> 
> 	cairo_set_source_surface (cr, x, y);
> 	cairo_paint (cr);

>    Currently, the first test passes, but the second test fails. The
>    problem here is some confusing semantics about the timing that
>    determines which user space to use for the user-to-pattern-space
>    matrix within the source pattern. (I had wanted the cairo_scale
>    call to scale the pattern, but not to scale the offset passed to
>    cairo_set_source_surface).

Maybe cairo_paint() should take the x,y arguments instead?