[Python-Dev] Review needed: proposed fix for 2.6 __hash__ incompatibility (issue 2235)

[Nick Coghlan]

I've posted a possible fix for the __hash__ backwards incompatibilities 
described in issue 2235 [1].

The patch uses a model similar to that used in Py3k (using None is 
indicate "don't inherit __hash__"), but extends it to allowing Py_None 
to be explicitly stored in the tp_hash slot. The major downside is that 
we suffer the cost of an extra pointer comparison on every call to 
PyObject_Hash, but I wasn't able to come up with another solution that 
preserved backwards compatibility while still allowing 
collections.Hashable to function correctly.

The patch involves a few changes to fairly deep components in 
typeobject.c though, so I'd like at least some kind of sanity check 
before I commit it.

Cheers,
Nick.

[1] http://bugs.python.org/issue2235
-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia
---------------------------------------------------------------
             http://www.boredomandlaziness.org