It would be handy to have "JavaSrc" encoders. This would let us remove
jv-convert from gcj. Also it would make native2ascii a bit less ugly.
Mark asked on irc so I thought I would expand.
A JavaSrc encoder applies a transform mentioned in the JLS.
It turns non-ascii characters into the sequence \uNNNN.
If it sees a literal \uNNNN in the input stream, this is turned
into \\uNNNN. (And likewise for any number of \s)
The decoder goes the other way, turning \uNNNN into a unicode
character and stripping a \ from other forms.
IOW it is an encoder that handles what native2ascii is currently
doing in plain old code.
Tom, how do you see this being implemented? As a character set encoder?