Martin,

I can almost agree with you _if_ I was setting the Content-Transfer-
Encoding myself, however I am not. I am setting the charset and the 
library chooses an appropriate Content-Transfer-Encoding to represent 
the mime part with. Currently I can't see any way other than reading 
the source or writing a test case (and that would require understanding 
what the email.mime module was doing "under the hood") for a developer 
to find out which Content-Transfer-Encoding was going to be used.

Also, just from a usability point of view I would expect that creating 
an invalid mime part would be a little more difficult. Especially 
considering the fix should be as small as adding "if not encoding in 
valid encodings: raise SensibleException".