Le Sam 2 avril 2011 02:18, Kevin Kofler a écrit :
Behdad Esfahbod wrote:
> To be honest, I don't fully agree with upstream. Autohinter and bytecode
> *can* be used together. The results won't be optimal, but much better
> than the current output. Maybe someone can bring up the multi-script font
> use case with upstream and convince them to reconsider.
Well, I think somebody would need to come up with a patch, too.
The current Freetype code is such that the decision whether to autohint is
taken very early, much before the glyph is even loaded at all. So there are
also technical obstacles to implementing per-glyph fallback.
Unlike upstream, I do think that the fallback should be per glyph, not per
font, but I don't know how to implement this.
IMHO, Fedora should use the hinter setting most favourable to the improvement
of FLOSS fonts. That's what Fedora mission *is*.
Activating bytecode is clearly a win for fully hinted small ASCII fonts and or
proprietary fonts. Because hinting (done properly) is hard, and FLOSS requires
releasing early, and often, instead of sitting on a font till its is
"complete", I suspect the current bytecode activation will only lead to :
1. stripping of partial hints from existing fonts (killing work and making
sure full hinting will never happen)
2. use of autohinting in font editors (for a result that can be worse than
freetype's autohinter, not sure if that permits manual hinting later or not)
Thus I tend to think like Behdad its current activation is a mistake. However
I don't create fonts myself, therefore I'd like to know what the opinion of
the people managing big FLOSS projects (Dejavu, Linux Libertine, etc) is on
this situation. And I'd defer to *their* opinion. The current change has
clearly been done just because "OMG a patent has expired it must enable good
stuff" and "Arial wants bytecode" not any serious analysis of what it
meant
from a FLOSS POW (some of the same people were staunch supporters of Luxi,
despite its legal status)
--
Nicolas Mailhot