AFM parsing and unencoded characters

[Ludovic Courtès]

Hi,

I'm again hitting the "ignoring unencoded ligature character" message.
However, this time, it yields a "fatal error: too many error messages"
because there are too many of them in the fonts at hand [0], which is
quite annoying.

I first thought about removing the error message limit, but that seemed
quite unappealing.  ;-)

Then I went back to z37.c and z38.c and found out that glyphs
("character metrics" would be more appropriately called "glyph metrics")
are referred to by an 8-bit integer (returned by `MapCharEncoding ()').

It would be wonderful if we could use long integers to identify glyphs
within a font.  AFM "unencoded characters" could be assigned any number
above 255[*].  The issue is that a lot of places currently assume 8-bit
integers (e.g., the `lig_table' field of `font_rec' is a 256-element
array).

Unfortunately, I probably won't have time to work on this, but I'd
appreciate comments and support.

Thanks,
Ludovic.


[*] However, the AFM specs don't explicitly restrict character codes to
    bytes.  Section 8 even contains an example using `CH' with a large
    character code, so this strategy may fail.

[0] http://linuxlibertine.sourceforge.net/