Re: Python's method-resolution algorithm: An implementation question

[Evan Aad]

Thanks!

On Tue, Aug 15, 2017 at 10:41 PM, Ian Kelly  wrote:
>
> On Tue, Aug 15, 2017 at 12:57 PM, Evan Aad  wrote:
> > I don't see how, since the L(B*)'s are listed in order in the argument
> > list: L(B1), L(B2), ..., and each L(B*) starts with B*: L(B1) =  > ...>, L(B2) = , ...
> >
> > Could you please give a counter-example?
>
> Sure.
>
> merge(, )  ->  
>
> vs:
>
> merge(, , )  ->  TypeError
>
> Or in Python:
>
> >>> class A1: pass
> >>> class A2: pass
> >>> class B1(A1): pass
> >>> class B2(A2, B1): pass
> >>> class C(B1, B2): pass
> Traceback (most recent call last):
>   File "", line 1, in 
> TypeError: Cannot create a consistent method resolution order (MRO)
> for bases B1, B2
> --
> https://mail.python.org/mailman/listinfo/python-list
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Python's method-resolution algorithm: An implementation question

[Ian Kelly]

On Tue, Aug 15, 2017 at 12:57 PM, Evan Aad  wrote:
> I don't see how, since the L(B*)'s are listed in order in the argument
> list: L(B1), L(B2), ..., and each L(B*) starts with B*: L(B1) =  ...>, L(B2) = , ...
>
> Could you please give a counter-example?

Sure.

merge(, )  ->  

vs:

merge(, , )  ->  TypeError

Or in Python:

>>> class A1: pass
>>> class A2: pass
>>> class B1(A1): pass
>>> class B2(A2, B1): pass
>>> class C(B1, B2): pass
Traceback (most recent call last):
  File "", line 1, in 
TypeError: Cannot create a consistent method resolution order (MRO)
for bases B1, B2
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Python's method-resolution algorithm: An implementation question

[Evan Aad]

I don't see how, since the L(B*)'s are listed in order in the argument
list: L(B1), L(B2), ..., and each L(B*) starts with B*: L(B1) = , L(B2) = , ...

Could you please give a counter-example?

On Tue, Aug 15, 2017 at 9:44 PM, Ian Kelly  wrote:
>
> On Tue, Aug 15, 2017 at 9:56 AM, Evan Aad  wrote:
> > According to the description of Python's method resolution order (mro)
> > (https://www.python.org/download/releases/2.3/mro/), a.k.a. C3
> > linearization (see Wikipedia), the algorithm can be described as
> > follows:
> >
> > "the linearization of C is the sum of C plus the merge of the
> > linearizations of the parents and the list of the parents."
> >
> > Symbolically, the algorithm is defined recursively thus:
> >
> > L(O) = 
> > L(C) =  + merge(L(B1),..., L(Bn), )
> >
> > where
> >
> > * O is the class from which every class inherits.
> > * C is a class that inherits directly from B1, ..., Bn, in this order.
> > * < and > are list delimiters.
> > * + is the list-concatenation operator.
> > * 'merge' merges its list arguments into a single list with no
> > repetitions in the manner described in the next paragraph.
> >
> > Consider the head of the first list, i.e L(B1)[0]: if it is a good
> > head, i.e. if it is not in the proper tail of any of the other lists,
> > add it to the linearization of C, and remove it from all the lists in
> > the merge. Otherwise, consider the head of the next list, etc. Repeat
> > until no more classes, or no good heads. In the latter case, it is
> > impossible to construct the merge.
> >
> >
> >
> > Why is the list  of parents given as an argument to
> > 'merge'? Will the algorithm produce different results if this argument
> > is omitted?
>
> It ensures that the parents are linearized in that order. Without that
> list, it would be possible for B2 to appear before B1 in the MRO of C.
> --
> https://mail.python.org/mailman/listinfo/python-list
-- 
https://mail.python.org/mailman/listinfo/python-list

Re: Python's method-resolution algorithm: An implementation question

[Ian Kelly]

On Tue, Aug 15, 2017 at 9:56 AM, Evan Aad  wrote:
> According to the description of Python's method resolution order (mro)
> (https://www.python.org/download/releases/2.3/mro/), a.k.a. C3
> linearization (see Wikipedia), the algorithm can be described as
> follows:
>
> "the linearization of C is the sum of C plus the merge of the
> linearizations of the parents and the list of the parents."
>
> Symbolically, the algorithm is defined recursively thus:
>
> L(O) = 
> L(C) =  + merge(L(B1),..., L(Bn), )
>
> where
>
> * O is the class from which every class inherits.
> * C is a class that inherits directly from B1, ..., Bn, in this order.
> * < and > are list delimiters.
> * + is the list-concatenation operator.
> * 'merge' merges its list arguments into a single list with no
> repetitions in the manner described in the next paragraph.
>
> Consider the head of the first list, i.e L(B1)[0]: if it is a good
> head, i.e. if it is not in the proper tail of any of the other lists,
> add it to the linearization of C, and remove it from all the lists in
> the merge. Otherwise, consider the head of the next list, etc. Repeat
> until no more classes, or no good heads. In the latter case, it is
> impossible to construct the merge.
>
>
>
> Why is the list  of parents given as an argument to
> 'merge'? Will the algorithm produce different results if this argument
> is omitted?

It ensures that the parents are linearized in that order. Without that
list, it would be possible for B2 to appear before B1 in the MRO of C.
-- 
https://mail.python.org/mailman/listinfo/python-list