We have a web service (closed source) that accesses an Oracle 11g database. It was recently upgraded from 10g to 11g. It returns records, one of the columns being an NCLOB type. The string sent to the webservice may contain Windows NewLines, \r\n. Unfortunately, I'm not sure what, if anything the web service was doing to manipulate the data sent/received from the DB.
In 10g, the string returned from the NCLOB column was Big Endian Unicode, and all '\r' were dropped, so new lines would return as a \n.
In 11g, the string returned from the NCLOB is ASCII encoded, and all '\r' were replaced with '\n', so new lines return as \n\n.
Does this seem reasonable? Honestly, we've been handling Oracle newline issues for a while (the behavior of 10g), and I'm pretty sure that this is a result of upgrading to 11g. Does anyone have information on differences between 10g and 11g, related to newline or escape character sequence storage or the NCLOB datatype? I'm trying to do damage control here and point the finger at Oracle 11g, but need some evidence.