new Buffer doesn't handle unicode well

[calvinmetcalf]

var arr = [255, 216, 255, 224, 0, 16, 74, 70, 73, 70, 0, 1]
//length 12;
var buf1 = new Buffer(arr)
// length 12
var hex1 = buf1.toString('hex');
// length 24
var str = buf1.toString('utf8')
// length 12
var buf2 = new Buffer(str);
//length 20
var hex2 = buf2.toString('hex');
// length 40;

results for hex2

browserify - efbfbd00104a4649460001 
node - efbfbdefbfbdefbfbdefbfbd00104a4649460001

the value of str in node is ����\u0000\u0010JFIF\u0000\u0001 which would suggest that the issue has to do with how new Buffer handles characters in text.

cc feross/buffer#66

[bnoordhuis]

255 is never valid in a UTF-8 byte sequence. It probably worked by accident but invalid UTF-8 is replaced with U+FFFD (the replacement character) as of node.js v0.10.29. The V8 UTF-8 decoder became stricter in v3.x so that might be a factor as well.

I don't think there is anything to do here except make a note of it somewhere in the documentation..

[bnoordhuis]

U+FFFD (the replacement character)

That's 0xEF 0xBF 0xBD as UTF-8, by the way; the repeated pattern you see in the test's output.

[calvinmetcalf]

fixed on the browserify end