gistfile1.txt

~ ⚡ node /tmp/readself.js 
length=214
// start of file readself.js
sys=require("sys");
fs=require("fs");
contents=fs.readFileSync(__filename, encoding="utf8"); // ö日本語
sys.puts("length="+contents.length);
sys.puts(contents);
// end of file readself.js


~ ⚡ wc -c /tmp/readself.js 
     221 /tmp/readself.js

Oh, and:

~ ⚡ file /tmp/readself.js 
/tmp/readself.js: UTF-8 Unicode C++ program text

Note how Node.js reads the file as UTF-8 and reports the length to be 214 bytes.
The file on disk is 221 bytes, a difference of 7.

The following shows that "// ö日本語" as a String has a length of 7.

sys=require('sys');
s="// ö日本語";
sys.puts(s)
sys.puts(s.length)

Remove the comment // ö日本語" and it works as expected.

What am I missing?

sure "length" is # characters not bytes but help me figure out what's up

"ö日本語" is four characters, and in UTF-8 "ö" is two bytes and the Han characters are each 3 bytes, so you have 7 bytes more than characters.