> This is the output of
> \markup \number 2
> using the official 2.25.2 Linux binaries on the one hand, and a
> self-compiled build made on the v2.25.2 tag from a clean directory
> on the other hand (PS backend in both cases).
My self-compiled LilyPond produces the 'other' glyph version (i.e.,
not `two.png`).
> If you look carefully at the tails on the right, you can see that
> they are different. Also, the first one has a thinner "choke point"
> on the left. This is easier to see on the attached screencast. Does
> anybody have an idea what might be causing this? Maybe different
> versions of METAFONT or FontForge?
Attached you can find the METAFONT proof sheet for glyph 'two'. Since
METAFONT is not involved in our font production chain (METAPOST is),
it is a good indicator that the 'other' glyph version is the right
one.
BTW, a PFB font without FontForge postprocessing can be created with
```
FONTFORGE=foo perl ../scripts/build/mf2pt1.pl --rounding=0.0001
feta-alphabet20.mf
```
For glyph 'two', the postprocessing effects are very subtle; FontForge
adds points at all extrema and reduces the number of points to get
smaller fonts (ensuring that the outline changes are less than a
certain threshold). I suspect that this last step doesn't work
correctly.
Werner
