Message266710
> Please treat this as a new feature (just in case) and only
> apply it to 3.6.
How about changing PyType_CheckExact to PyType_Check for 2.7 and 3.5? It solves the original problem by expanding the single-argument case to metaclasses that aren't an exact instance of `type`. I think that's unlikely to break existing code. It also shouldn't cause a performance problem since the PyType_Check macro uses the Py_TPFLAGS_TYPE_SUBCLASS flag. |
|
Date |
User |
Action |
Args |
2016-05-30 17:31:19 | eryksun | unlink | issue27157 messages |
2016-05-30 17:23:42 | eryksun | set | recipients:
+ eryksun, gvanrossum, rhettinger, steven.daprano, r.david.murray, serhiy.storchaka, ppperry, abarry |
2016-05-30 17:23:42 | eryksun | set | messageid: <1464629022.51.0.277958806273.issue27157@psf.upfronthosting.co.za> |
2016-05-30 17:23:42 | eryksun | link | issue27157 messages |
2016-05-30 17:23:42 | eryksun | create | |
|