Linear translations are not deleted by the parser, regardless of whether they are regular characters or in the CDATA section. But in both cases, various line feeds (Unix, Windows, Mac) are normalized to a single-character canonical ("unix", \ n) line feed. It is impossible to prevent this normalization, with the exception of using a character entity as suggested; and this cannot be done in the CDATA section because data processing is disabled there.
But why exactly do you want to prevent this normalization? If you want this to display, you can simply replace \ n with any local translation line (\ r for mac or \ r \ n for windows).
source share