Re: [Devel] RE: lack of a Unicode character map should not prevent the typeface from being loaded

[David Turner]

Hello,

Graham Asher wrote:

> I said earlier today:
>
> <<<<<<
> In doing some more work to the GhostScript-to-FreeType bridge I have
> discovered a small bug, probably a recent regression. The function open_face
> in ftobjs.c should not fail if no Unicode character map is found. The fix is
> simple:
>  
Thanks a lot Graham, I've commited your fix. By the way, I've looked at 
Antoine's work and would
like to thank him as well. I have recently added the following comment 
in "find_unicode_charmap" to
make things clear to people not too used to the TrueType specifications:

   /*
    *  the original TrueType specification(s) only specified charmap
    *  formats that are capable of mapping 8 or 16 bit character codes to
    *  glyph indices.
    *
    *  however, recent updates to the Apple and OpenType specifications
    *  introduced new formats that are capable of mapping 32-bit character
    *  codes as well. And these are already used on some fonts, mainly to
    *  map non-BMP asian ideographs as defined in Unicode.
    *
    *  for compatibility purposes, these fonts generally come with
    *  *several* Unicode charmaps:
    *
    *   - one of them in the "old" 16-bit format, that cannot access
    *     all glyphs in the font
    *
    *   - another one in the "new" 32-bit format, that can access all
    *     the glyphs.
    *
    *  this function has been written to always favor a 32-bit charmap
    *  when found. Otherwise, a 16-bit one is returned when found
    */

Cheers,

- David Turner
- The FreeType Project  (www.freetype.org)