Comment 1 for bug 1463609

Also this:

>>> docsrc = u'\ufeff<?xml version="1.0" encoding="us-ascii"?><root/>'.encode('utf-8')
>>> lxml.etree.fromstring(docsrc)
<Element root at 0x10c7c1ef0>

Because of the byte order marker '\ufeff', the document is not ascii.