[Python-Dev] perplexed by mro

[Samuele Pedroni]

more on 2.2 algo:

> > >>> cpl.ex5(globals())
> > class a(object): pass
> > class b(object): pass
> > class c(object): pass
> > class x(a): pass
> > class y(a): pass
> > class z(x,b,y,c): pass
> > >>> cpl.names(z.__mro__) # just the class names, real 2.2 mro algo
> > ['z', 'x', 'y', 'a', 'b', 'c', 'object']

now b is there, but a simple substitution will change things radically:

>>> cpl.ex5perturb(globals())
class a(object): pass
class b(object): pass
class c(object): pass
class x(a): pass
class y(a): pass
class y_b(b,y): pass
class z(x,y_b,c): pass
>>> cpl.names(z.__mro__)
['z', 'x', 'y_b', 'b', 'y', 'a', 'c', 'object']

instead of inheriting directly from b and y, z inherits from y_b that inherits
from both. Now b is between x and a.
> > >>> cpl.ex9(globals())
> > class a(object): pass
> > class b(object): pass
> > class c(object): pass
> > class d(object): pass
> > class e(object): pass
> > class k1(a,b,c): pass
> > class k2(d,b,e): pass
> > class k3(d,a): pass
> > class z(k1,k2,k3): pass
> > >>> cpl.names(z.__mro__)
> > ['z', 'k1', 'k3', 'a', 'k2', 'd', 'b', 'c', 'e', 'object']
>
> I think this one *does* have an order conflict, and in that case all
> bets are off about the book's algo.  If you look at k1 and k2, clearly
> d must come after a; yet k3 wants it after a.
>

yes but for class z(k3,k1,k2) the conflict would go away. It's not intuitive
nor stable.

regards.