The perfect solution for the UTF-8/16 discussion

[Markus Scherer]

Abolish all in-process Unicode encodings except UTF-16.
If everyone uses the same encoding form then there is no problem with different string 
lengths, results of binary comparisons, etc.

Once we are here, abolish all little-endian UTF-16 implementations. This will save a 
lot of byte swapping, and binary comparisons can always be performed with memcmp().

Heck, abolish all little-endian platforms and all platforms with integer widths other 
than 8, 16, 32, etc.

:-)

markus