Should multi-line raw strings preserve line encoding?

[DartBot]

This issue was originally filed by zerny...@google.com

---

Running:

main() {
  String s1 = 'foo\r\n';
  String s2 = r'''foo^M
''';
  print(s1.codeUnits);
  print(s2.codeUnits);
}

Will output:
[102, 111, 111, 13, 10]
[102, 111, 111, 10]

Before running, replace ^M by an actual carriage return. In Emacs insert a carriage return with C-q RET
Compiling with dart2js and running with d8 produces the expected output:
[102, 111, 111, 13, 10]
[102, 111, 111, 13, 10]

[mhausner]

That would mean that a raw string can change depending on which platform you save the source code. Save it on Windows and you get \r\n on every line ending. Save the same file on Linux and you'll get a string with \n at the end of every line. That seems less than ideal.

[mhausner]

Gilad, could you clarify whether raw strings have to preserve the platform-specific line ending encoding? The spec does not mention this issue other than stating that source is in UTF-8, which leaves line endings platform-dependent. For raw strings, the spec just says that escape sequences and interpolations are not processed.

Thank you.