Seems that in python3 the XML encoding declaration from xml.etree.ElementTree has changed from 2.x in that it is now lowercased, e.g. 'utf-8'. While the XML spec [1] says that decoders _SHOULD_ understand this, the encoding string _SHOULD_ be 'UTF-8'. It seems that keeping to the standard in the vein of being strictly conformant in encoding, lax in decoding will give maximum compatibility.

It also seems like an unhelpful change for 2.x to 3.x migration though that is perhaps a minor issue (but how I noticed it).

Can show with:

>cat a.py
from xml.etree.ElementTree import ElementTree, Element
import os, sys
print(sys.version_info)
if sys.version_info > (3, 0):
    fp = os.fdopen(sys.stdout.fileno(), 'wb')
else:
    fp = sys.stdout
root = Element('hello',{'beer':'good'})
ElementTree(root).write(fp, encoding='UTF-8', xml_declaration=True)
fp.write(b"\n")

>python a.py
sys.version_info(major=2, minor=7, micro=5, releaselevel='final', serial=0)
<?xml version='1.0' encoding='UTF-8'?>
<hello beer="good" />

>python3 a.py
sys.version_info(major=3, minor=4, micro=2, releaselevel='final', serial=0)
<?xml version='1.0' encoding='utf-8'?>
<hello beer="good" />

Cheers,
Simeon

[1] <http://www.w3.org/TR/2006/REC-xml11-20060816/#NT-EncName> "In an encoding declaration, the values "UTF-8", "UTF-16", ... should be used for the various encodings and transformations of Unicode" and then later "XML processors should match character encoding names in a case-insensitive way".