Say we parse a file and generate two infosets, one via -I xml and -I jdom. These two infosets are functionally the same, except that the one generated via -I xml has newlines and indentation. If we unparse using -I jdom with each of the different files, the one generated via -I xml is significantly slower than the one generated with -I jdom. This implies that there is a noticeable performance hit with whitespace in a JDOM tree. We should investigate if this is that case and determine a fix.