Re: [ft-devel] GETVARIATION and GETDATA

[Werner LEMBERG]

[You have a really strange citation style.  What e-mail program are
you using?]

>> And now I remember again: The font is OK, there are *no*
>> undocumented instructions at all!  It properly uses IDEF
>> everywhere.  While unusual, this is fully valid bytecode, even on
>> MS engines.  In other words, Microsoft removed support for
>> GETVARIATION and GETDATA in its bytecode engine but forgot to
>> update the validator program.
>  
> Argh - I think I understand it a lot better than I did a year ago
> now.  It is not that the two font uses Apple-specific instructions,
> but that it does things Microsoft-specific way, and depends on the
> rendering engine being non-Apple. :-).

Exactly.

> Same thing - I think a proper description of the problem is that
> "these font use instructions which have different meanings on
> different rendering engines".

No.  The correct description is rather

  IDEFs for bytecodes 0x91 and 0x92 are not possible with the Apple
  bytecode engine since those instructions are undocumented but still
  in use for backwards compatibility.

> The validator program does flag Skia.ttf having Apple-specific
> instructions also, but that's fully expected that Apple ships fonts
> with apple-specific instructions.

Handling 0x91 on both engines is easy: It is enabled for GX glyph
variation fonts only.

...  OK, I've changed my mind.  Instead of removing 0x92 again, I have
made this instruction active for GX fonts only so that an IDEF for
0x92 is still possible with a standard TTF.


    Werner