Python 3.2a0 (py3k:78205M, Feb 16 2010, 17:32:08)
[GCC 4.4.1] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> e = None
[50279 refs]
>>> e
[50279 refs]
>>> try: raise ValueError
... except ValueError as e: pass
...
[50277 refs]
>>> e
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'e' is not defined
[50277 refs]


Same for 3.1.1. Note how the refcount is going down after the try-except.

This is a not a bug and it's documented: http://docs.python.org/py3k/reference/compound_stmts.html#the-try-statement

The solution is to assign the value to another variable:
>>> try: raise ValueError
... except ValueError as e: exc = e
...
>>> exc
ValueError()
>>> e
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'e' is not defined

I knew that the variable was supposed to go out of scope when leaving the except block, but I wasn't aware that this was supposed to be done using (the equivalent of) a 'del'. Were the side-effects of deleting variables defined outside of the try-except considered negligible at the time? I can't see a warning about that in the docs.

In this case I don't see much difference between deleting a variable or assigning it to something else.

This code works on both Python 2 and 3:
>>> e = 'test'
>>> try: pass                    # no errors raised here
... except Exception as e: pass  # this is not executed
...
>>> e
'test'

If you do this instead:
>>> e = 'test'
>>> try: raise ValueError        # raise the error
... except Exception as e: pass  # this is executed
...

On both 2.x and 3.x 'e' doesn't refer to 'test' anymore in the moment that the exception is captured by the except, so what happens next to the "new e" is not related to what the "old e" was.

See also #4617, with some patch.