Jaysinh, unless a patch for 3.5 does not cleanly apply to 3.6, or the patch must be different in 3.6*, no 3.6 patch is needed.  Ditto for 3.7/default. Mariatta's patch merged forward without problem. *In this case, 'must be different' would be because there are different code examples.

The same is true for 2.7 if the 3.x patch applies cleanly there (though this is less common there).  Mariatta's patch also applied to 2.7, though as I reported, I found one more change for 2.7.

So when you revise in response to comments, one of each patch is enough unless you know more are needed, in which case say so.  If there are problems forward-merging the other_rst patch, please report first before preparing a 3.6 patch.  If there were a problem with just one of the files, I would prpbably want a 3.6 patch just for the one file.

I discovered a deeper problem.  In spite of the doctest annotations, the doctests is not run on the ctypes doc and at least one of the tracebacks is wrong.

>>> windll.kernel32.GetModuleHandleA()      # doctest: +WINDOWS	  Traceback (most recent call last):	
  File "<stdin>", line 1, in ?	
    ValueError: Procedure probably called with not enough arguments (4 bytes missing)
>>> windll.kernel32.GetModuleHandleA(0, 0)  # doctest: +WINDOWS
Traceback (most recent call last):
  File "<stdin>", line 1, in ?	   ValueError: Procedure probably called with too many arguments (4 bytes in excess)

whereas, on Win10, 3.5
>>> from ctypes import *
>>> windll.kernel32.GetModuleHandleA()
0
>>> windll.kernel32.GetModuleHandleA(0, 0)
480968704

However, when I ran make doctest, these did not show up, but they were not run.  I will worry about this as a separate issue.

>>> windll.kernel32.GetModuleHandleA()
477822976
>>> windll.kernel32.GetModuleHandleA(0, 0)
477822976