The "bdf2psf" utility that is used to generate Linux console fonts
expects an "equivalences" file as argument and uses this file to
determine characters that should share the same glyph. The file
"unicode.subst" lists possible replacement characters for characters
that are not contained in some encoding, but it provides a preference
list, not an equivalence list: An entry "A B C" indicates that if
character A is not present, it should be replaced preferably by B and
at a pinch by C. It does not indicate that A, B, and C look the same,
or that A should be replaced by B even if A is present, or that A or B
are suitable replacements for C, or that A or C are suitable
replacements for B. For any given encoding, "bdftoequiv" converts
"unicode.subst" into an "equ" file that can be used by "bdf2psf".

Some of the replacements in "unicode.subst" depend on the specific
shapes of glyphs in UW ttyp0. They may not look good with other fonts.
Similarly, replacements that are suitable for other fonts may not be
appropriate for UW ttyp0.

The Linux console does not handle combining accents properly. Instead
of printing combining accents above or below the base letter, it
prints the combining accent after the base letter. Some of the
replacements in "unicode.subst" exploit this deficiency; they would
be unsuited if the Linux console were able of overprinting.
