Re: [Devel] Apple's 'gvar' table and friends

[George Williams]

On Tue, 2004-03-23 at 23:57, Werner LEMBERG wrote:
> > The gxlayout contains the loader for fvar.
> > The max of coordinates(== 4) is statically given  in the multi-master
> > interface of FT2.
> 
> Please advise how to intergrate best -- since gvar is very tightly
> connected to the rasterizer we need a low-level interface in case
> gvar should be handled by a higher level.
I should not think there would be any need for a higher level interface
to 'gvar', any more than there needs to be a high level interface to
'loca'.

As Werner says, both 'gvar' and 'cvar' need to be done at a low level, I
don't think relying on gxlayout is appropriate. Especially since, as I
understand things, it's going to be a layer on top of freetype?

'fvar' needs both a low-level and a high-level interface (or perhaps I
should say an internal and external (user) interface), but the internal
needs are different from those of a higher level. The internal routines
need to know how many axes there are and what the bounds are on those
axes. They also need access to the data in the 'avar' table (if
present). External routines need the name mappings and the bounds, but
don't need the 'avar' data.

On Tue, 2004-03-23 at 23:20, Werner LEMBERG wrote: 
> Yes, definitely.  What kind of interface do you suggest?
As I see it the interface to the user (at the freetype level) should
probably be fairly simple. I only see two things as needed:
   * An interface to get the contents of 'fvar'
        This would provide the data that gxlayout provides, but it
        still seems to me that gxlayout is not the right way to
        find these data...
   * An interface to set the location on the axes
        (to be called before Set_Pixel_Sizes)
There might also be a patch to Set_Pixel_Sizes/Set_Char_Size so that it
would automagically find the right spot on the Optical Size axis.