From: Hans Aberg (haberg@math.su.se)
Date: Wed Jun 17 2009 - 14:02:38 CDT
It is possible to make endian independent 32-bit encodings, since the
number of equivalences classes under endian swapping (2^31 + 2^15 =
2147516416) is (much) larger than the number of code points, though at
some expense on encoding/decoding overhead. I made one such encoding,
chosen to make encoding/decoding easy (C++ code attached; details in
the file). The lower 0x7FFF code points are mapped (modulo endian
swapping) to identity, and encoded characters need not be fully
decoded for comparison of code points.
Hans Aberg
This archive was generated by hypermail 2.1.5 : Wed Jun 17 2009 - 14:05:08 CDT