Carriage returns get filtered out when loading files, so the in memory representation always only has line feeds. And when saving it just save the in memory representation (so UTF-8 and with line feeds). I haven't added support for saving it with the original encoding and line endings because I never needed it, but yeah that should be easy to add. But for the internal representation I would probably still only use line feeds because supporting carriage returns in there would make some things more complex