Some deprecated ElementTree features are deprecated only in the documentation or in Python implementation (that is virtually the same since C implementation is default). Proposed patch adds missed deprecations is code. It also makes warnings be ignored only in tests where they are expected. This is possible since converting doctests to unittests some time ago.
Element.getchildren() and Element.getiterator() methods. They were deprecated in the documentation and in Python implementation in 2.7 and 3.2.
The xml.etree.cElementTree module. Deprecated in the documentation in 3.3.
The html argument of XMLParser. Deprecated in the documentation in 3.4.
Ned, is it appropriate to commit the patch (or its part) in 3.6? The discrepancy between Python and C implementation can be considered as a bug. What are your thoughts?
Regarding the cElementTree module, this is a bit problematic. The idiomatic import has lost its use in Py2.5 when ET and cET were added to the stdlib, so code that was written for Py2.5 or later (e.g. because it uses generators) might no longer have that cascade. On the other hand, issuing a warning for the module would also hit this import cascade, even though the code would work just fine without cElementTree. One argument speaks for deprecation, the other for removal.
However, cElementTree is redundant now, so it should be removed eventually. And since that removal would break some code anyway, I'd be ok with just removing it without prior import warnings. People can then decide whether they want to fix their code by adding the well-known import cascade (and not get annoying warnings for it) or by switching entirely to plain ET and not looking back.