getting closer

i can now compile and run with unicode-internal and even load up etc/HELLO, where i duly see "zdravstvuite" in cyrillic and "geia sas" in greek. the three versions of the first char after GB/JIS/KSC all report the same char values, which is a good sign. the big5 chars appear as ~; this is because they are encoded using the fake big5-1/big5-2 charsets, which unicode-internal knows nothing about. probably i should redo etc/HELLO to use extended segments to encode big5. old-mule has improvements, too; it knows how to handle arbitrary unicode chars (at least, anything up through 0x31FFF) and allows more charsets than before -- 96 private in dimension 1 and another 96 private in dimension 2. more testing needed before it will be committed and font-handling in unicode-internal still needs an overhaul. ben