The font data internally starts at glyph 0 and goes up
through N. For some fonts, they may not begin at glyph
0, and may also have large internal gaps.
gleanfonts somewhat compensates for this by using a
"filler" character (which should be font-dependant), but
it would be better if somehow one could record the
strips of used glyphs, rather than storing a duplicate
reference to some filler character in every "unused"
Implementing this would change the genfonts program as
well as font data files and would require the
text rendering routines to change.
This is not a terribly urgent matter, as the filler
characters currently require "just" a pointer.
On the other hand, if one could drop the unused symbols
altogether, then the used symbols could be directly
represented (possibly a strip of glyphs being a single
bitmap), allowing more efficient storage and perhaps
more efficient access.