Re: [DejaVu-fonts] worse the vera, reloaded

[<pc...@gm...>]

Hi Keenan,

> > The first one (http://dejavu.sourceforge.net/wiki/images/3/33/Dejavu_no=
tsogood.png)
> > is probably caused by some undocumented behavior of the SHZ
> > instruction which freetype does not implement. I'll better doublecheck
> > everything this time before coming up with any claims about this :-)
>=20
> By far the majority of Vera's DELTAP1 instructions involve the n+1 advanc=
e
> point; maybe we could fix the spacing problems just by tweaking those.
>=20
> For example, the attached patch fixes the problem with lowercase z shown =
in the
> image by simply removing one of the DELTAP1 instructions, so the advance =
point
> remains where it is instead of being shifted one pixel to the left.

Hehe, nice try. Check out
http://sourceforge.net/mailarchive/forum.php?thread_id=3D7908822&forum_id=
=3D40874,
third message from top... ;-)

This does not work though, as it will make 'z' too wide on cleartype
and maybe on other renderers. The wrong widths always appear after a
certain sequence of instructions (look for the mppem - if - shpix -
shz - shpix - endif sequence). I'll post more details if anyone's
interested.

> > The second one (http://dejavu.sourceforge.net/wiki/images/e/ea/Ttf_kern=
ing_bug.png)
> > is pretty easy. Fontforge does not set the right flags (again). Take
> > for example the 'scaron' glyph. The instructions in the 's' component
> > change the advance width of this component, but the composite 'scaron'
> > does not "inherit" the new width and is scaled linearly. This is
> > controlled by the USE_MY_METRICS flag (see [...]

> It sure would be nice to have support in FontForge for this stuff. I thin=
k I'll
> subscribe to the fontforge-devel mailing list and bug them, err... polite=
ly
> request for this to be implemented. =3D)

It surely is worth a try. There is even no need to change fontforge's
GUI, just preserving the flag in the sfd files would be sufficient.
Good luck with ff's devs then.

-- Peter