Re: Type 2 16.16 fixed numbers

[David Turner]

> 
> Just an update on the 16.16 fixed number stuff in t2gload.
> 
> If I am not mistaken, the reason why Adobe hijacked the 5 byte
> encoding to get a 16.16 fixed number is that now one can get a
> fractional number without having to resort to div (as one has to do
> within a Type 1 font).
> 
> So, shifting right 16 in add_point is probably a bad thing to do, as
> it clobbers the fractional part (David's argument about .05%
> difference not withstanding).  Then again, not shifting is also
> probably a bad thing to do, because of the format the rasterizer
> expects the numbers to be in.
> 
> Oh well...
> 

I believe that having the 16.16 format internally was important
for Type 2 in order to support interpolations with high accuracy
when dealing with Multiple Master fonts, not when rendering the
glyph themselves.

Since MM is now dead in CFF, it's probably more of a nuisance than
anything else now. However, I'd tend to think that the current code
seems to work, we don't necessarily need to "outsmart" the
specification, so let's keep it as is, it's not such a problem
anyway..

Cheers,

- David