but looks like that was a problem before this change.

Closes patch #1500623.  Thanks, Collin Winter.

functools, and adjust docs etc.

traceback.py be closer to the built-in formatting.
A few unittests had to be fixed, too.

Don't use our own temp name creation (dbm).
Should be backported to 2.5.

Create a real zip iterator object; not using itertools.izip
(Brian Holmes).

import MAGIC number.

``python -m compileall -f`` is really handy for finding Python files that use
invalid syntax.

Still need to remove traces of the UNARY_CONVERT opcode.

optimization for it.

SyntaxError.

This is probably not the proper solution to this failing test, but removing the
test itself causes 19 other tests to fail for some odd reason because doctest
doesn't expect a complete traceback (or something; rather odd problem for just
removing a single test).

auto-generates this file will need to get fixed.

'<>'.

use __eq__ instead of __cmp__.  The other change is unexplained:
with a random hash code as before, it would run forever; with a constant
hash code, it fails quickly.

This found a refcount bug in dict_equal() -- I wonder if that bug is
also present in 2.5...

I mea, *really* equal -- for now, the implementation just imports
itertools. :-)
The only other changes necessary were various unit tests that were
assuming zip() returns a real list.  No "real" code made this assumption.

This essentially meant fixing one case where a list of custom objects
was being sorted, and fixing one genuine bug where a method call was
missing parentheses.

The special-casing of other objects with a timetuple attribute is gone.
Let's hope Tim agrees.

This makes test_codeop and test_marshal pass.