Serhiy Storchaka:
> As for update_one_slot() see also issue5322 and issue25731.

Oh, thanks for the pointers! Now I understand much better these bugs.

I'm quite sure that they are still flaws in this code when a type is modified after PyType_Ready(), like sysmodule.c::

    /* prevent user from creating new instances */
    FlagsType.tp_init = NULL;
    FlagsType.tp_new = NULL;


But each time I had to dig into typeobject.c, my head is going to explode :-D I may try to understand one more time, but not today ;-)