Welcome to version 0.3 of FontAnvil.  It hasn't been very long,
chronologically, since the last release, but version 0.2 had a number of
serious bugs, and there have been some significant behind-the-scenes changes
as well, so a new release seems in order.  Several tens of thousands of
lines of code have changed between the two.

New in this version:  George's messy handrolled tokenizers for PE script and
the Adobe feature file language have been replaced by slightly less messy
lex-generated scanner code, which is no less complicated but I hope will be
more maintainable.  Similarly, most of the hardcoded multi-level tables used
for encoding translations have been replaced by auto-generated code from my
new (and still highly experimental) Icemap utility.  You will need Icemap to
compile FontAnvil from a source control checkout, but since Icemap is in the
same repository, that should be no big deal.

Many of those tables have been simply removed; it appears that the only
purpose they were serving was to allow script code to be written in native
8-bit encodings other than Unicode.  It is not clear to me that that's even
a good idea, since writing a script in an encoding other than ASCII and then
distributing it will make trouble for whoever receives a copy.  I certainly
think it's unnecessary for FontAnvil to support scripts written in the
proprietary encoding of Adobe's Zapf Dingbats font, so the hundreds of lines
of FontForge code that apparently existed only to support that case (!) have
been removed.

This version also includes a much improved test suite.  However, it may be a
long time before FontAnvil is actually able to pass its tests.
