Re: [ft-devel] units_per_em in CFF/CID FDArray

[Dirck Blaskey]

Werner LEMBERG wrote:

> After some investigation I think this is the wrong approach.  Note
> that each CID subfont can have a different FontMatrix value!  For the
> user, only the top dict's entry should be returned.
>
> The right approach IMHO is to select the subfont's FontMatrix in
> cff_slot_load, similar to the code in the CID module.
>
> Below is an untested patch which I ask you to verify (and fix if
> necessary :-).
>  

Hi Werner,

Thanks for taking a look at this.  I can see why you'd like to take the 
approach
you've outlined.  Unfortunately, it doesn't work.

It looks to me like in cff_parse_font_matrix, once it extracts the 
units_per_EM
value, it renormalizes the matrix back to 0x10000 0 0 0x10000 etc.
So fonts with 1000 or 2048 units_per_EM have the same matrix in the
cff->subfonts[fd_index]->font_dict.font_matrix where you're looking.

If you want to pursue it further, attached is the 2048.cff font.
Let me know if you need any more information.

thanks again,

d

2048.cff
Description: Binary data