DTTXML should not die upon reading "malformed" files.
When reading a file, DTTXML can raise IOExceptions or an exception if the file is not an XML file, but otherwise should complete the read and return to the user all data it was able to parse, ignoring the rest or giving warnings where appropriate.
Malformed sections or unexpected sections that are unreadable to DTTXML should not prevent the reading of some other correctly formed section.
See for example the tests of files with references, which currently fail. DTTXML cannot properly parse reference sections and raises an exception. There is no way for the user to continue the read and retrieve other data from the same file except manually load the file and delete the references. Instead, DTTXML should just print a warning and continue.