According to Taylor, in the test for handling double newline entities as separators, there's a discrepancy between the result returned via our Scala api vs the java api. We are translating from JDOM to Scala Node to JDOM. So it keeps replacing entities when it doesn't need to.
The infoset uses XML entities to represent '<' and '>' However, the ampersands in the entity are getting replaced. sobecomes instead of
See test doubleNL in the following files:
I created a version of the test that uses CDATA tags and literal '<' and '>' characters (doubleNL2) and this seems to be a good workaround.