Conflict Resolution using Maximum or Minimum Values

The zope.minmax.AbstractValue class provides a super class which can be subclassed to store arbitrary homogeneous values in a persistent storage and apply different conflict resolution policies.

class zope.minmax.AbstractValue(value=None)

Abstract implementation of zope.minmax.interfaces.IAbstractValue.

Subclasses must implement _p_resolveConflict.

The subclasses defined here are resolving the conflicts using always either the maximum or the minimum of the conflicting values.

Maximum

class zope.minmax.Maximum(value=None)

The zope.minmax.Maximum class always resolves conflicts favoring the maximum value. Let’s instantiate one object and verify that it satisfies the interface.

>>> import zope.minmax
>>> import zope.interface.verify
>>> max_favored = zope.minmax.Maximum()
>>> zope.interface.verify.verifyObject(
...     zope.minmax.interfaces.IAbstractValue, max_favored)
True

We can confirm that the initial value is zero.

>>> bool(max_favored)
False
>>> print(max_favored.value)
None

Now, we can store a new value in the object.

>>> max_favored.value = 11
>>> print(max_favored.value)
11
>>> bool(max_favored)
True

Or we can use the methods.

>>> max_favored.__setstate__(4532)
>>> max_favored.__getstate__()
4532
>>> print(max_favored.value)
4532
>>> bool(max_favored)
True

Do notice that using a direct assignment to the value attribute is a more natural use.

Minimum

class zope.minmax.Minimum(value=None)

The zope.minmax.Minimum class always resolves conflicts favoring the minimum value. Again, we instantiate an object and verify that it satisfies the interface.

>>> min_favored = zope.minmax.Minimum()
>>> zope.interface.verify.verifyObject(
...     zope.minmax.interfaces.IAbstractValue, min_favored)
True

We need a confirmation that the initial value is zero.

>>> bool(min_favored)
False
>>> print(min_favored.value)
None

Let’s populate this one too.

>>> min_favored.value = 22
>>> print(min_favored.value)
22
>>> bool(min_favored)
True

Or we can use the methods, again.

>>> min_favored.__setstate__(8796)
>>> min_favored.__getstate__()
8796
>>> print(min_favored.value)
8796
>>> bool(min_favored)
True

Please, notice, again, that using a direct assignment to the value attribute is a more natural use.

Conflict Resolution

Now, we need to exercise the conflict resolution interface. First for the zope.minmax.Maximum:

Let’s try differing values larger than the old value.

>>> max_favored._p_resolveConflict(max_favored.value, 4536, 4535)
4536
>>> max_favored._p_resolveConflict(max_favored.value, 4573, 4574)
4574

What happens when all the values are equal, including the old.

>>> max_favored._p_resolveConflict(max_favored.value, 4532, 4532)
4532

Notice that when the old value is larger than both the committed and new, it is still disregarded.

>>> max_favored._p_resolveConflict(max_favored.value, 4531, 4530)
4531

Now, the zope.minmax.Minimum:

Let’s try differing values smaller than the old value.

>>> min_favored._p_resolveConflict(min_favored.value, 8792, 8791)
8791
>>> min_favored._p_resolveConflict(min_favored.value, 8785, 8786)
8785

What happens when all the values are equal, including the old.

>>> min_favored._p_resolveConflict(min_favored.value, 8796, 8796)
8796

Notice that when the old value is smaller than both the committed and new, it is still disregarded.

>>> min_favored._p_resolveConflict(min_favored.value, 8798, 8799)
8798

How about an example that is not numerical?

>>> max_word = zope.minmax.Maximum('joy')
>>> print(max_word.value)
joy
>>> bool(max_word)
True
>>> max_word._p_resolveConflict(max_word.value, 'happiness', 'exuberance')
'happiness'
>>> max_word._p_resolveConflict(max_word.value, 'exuberance', 'happiness')
'happiness'
>>> min_word = zope.minmax.Minimum(max_word.value)
>>> print(min_word.value)
joy
>>> bool(min_word)
True
>>> min_word._p_resolveConflict(min_word.value, 'happiness', 'exuberance')
'exuberance'
>>> min_word._p_resolveConflict(min_word.value, 'exuberance', 'happiness')
'exuberance'

As indicated, we don’t need to have numbers, just homegeneous items. The homogeneous values are not really inherently required. However, it makes no sense to apply min() or max() on, say, one number and one string. Simply, the ordering relations do not work at all on heterogeneous values.