[issue19787] Fix PyThread_set_key_value() strange behaviour

[STINNER Victor]

STINNER Victor added the comment:

My commit broke test_capi, so I revert it. Here is the commit as a patch.

The impact on subinterpreters is more complex than what I expected. A decision should be take on what to do: mimic behaviour of Python 3.3 for subinterpreters (don't replace the Python thread state when a new subinterpreter is created), or fix code using subinterpreters.

----------
resolution: fixed -> 
status: closed -> open
Added file: http://bugs.python.org/file33110/pythread_set_key_value-2.patch

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue19787>
_______________________________________