Immutability and Python

Introducing Pyrsistent

Disclaimer

Some of the stuff mentioned here is specific for CPython.

For other interpreters other rules may apply.

What is immutability?

Something is immutable if it cannot be changed once created.

Lets keep it simple:

Why immutability in programming?

Cognitive offload for the developer
Safe and simple invariance checking
Safe and fast sharing, within and between threads
Safe and fast reuse
Hashable
Runtime efficiency

Why not?

Runtime efficiency
Habits
Language support

Immutability classes (WAT?)

Immutable by convention
Apparent immutability
True immutability

Immutability in Python

No constant references

Base types

>>> x = 17
>>> y = 17
>>> x is y
True
>>> y = 1700
>>> x = 1700
>>> x is y
False

Numbers and strings are immutable

Collections

tuple, frozenset and namedtuple, that's about it...

>>> tuple() is tuple()
True
>>> list() is list()
False

Gets messy and inefficient when you want to evolve the content

>>> t1 = (1, 2, 3, 4, 5)
>>> t2 = t1[:2] + tuple(17) + t1[3:]
# or
>>> l1 = list(t1)
>>> l1[2] = 17
>>> t2 = tuple(l1)

Enter Pyrsistent

Immutable collections that are easy to evolve

a.k.a. persistent data structures

a.k.a. functional data structures

Pyrsistent eco system

a = apparent immutability

c = immutability by convention

Examples

>>> students = freeze([{'name': 'Ann', 'grade': 4.1}, {'name': 'Bob', 'grade': 2.1}])
>>> students_v2 = students.set_in((1, 'grade'), 1.9)
>>> students
pvector([pmap({'grade': 4.1, 'name': 'Ann'}), pmap({'grade': 2.1, 'name': 'Bob'})])
>>> students_v2
pvector([pmap({'grade': 4.1, 'name': 'Ann'}), pmap({'grade': 1.9, 'name': 'Bob'})])
>>> thaw(students_v2)
[{'grade': 4.1, 'name': 'Ann'}, {'grade': 1.9, 'name': 'Bob'}]

>>> Point = pclass('x, y', name='Point')
>>> p = Point(1, 2)
>>> p2 = p.set(x=3)
>>> p
Point(x=1, y=2)
>>> p2
Point(x=3, y=2)

>>> v1 = pvector([1, 2, 3, 4])
>>> v1 == v(1, 2, 3, 4)
True
>>> v1[1]
2
>>> v1[1:3]
pvector([2, 3])
>>> v3 = v1.set(1, 5)
>>> v3
pvector([1, 5, 3, 4])
>>> v1
pvector([1, 2, 3, 4])
>>> pvector(x + 1 for x in v1)
pvector([2, 3, 4, 5])

>>> m1 = pmap({'a':1, 'b':2})
>>> m1 == m(a=1, b=2)
True
>>> m1['a']
1
>>> m1.b
2
>>> m1.items()
[('a', 1), ('b', 2)]
>>> m1.set('a', 3)
pmap({'a': 3, 'b': 2})
>>> m1
pmap({'a': 1, 'b': 2})

Structural reuse

http://hypirion.com/musings/understanding-persistent-vector-pt-1

>>> v1 = v(0, 1, 2, 3, 4, 5, 6, 7, 8)
>>> v1
pvector([0, 1, 2, 3, 4, 5, 6, 7, 8])
>>> v2 = v1.set(5, 'beef')
>>> v2
pvector([0, 1, 2, 3, 4, 'beef', 6, 7, 8])
>>> v1
pvector([0, 1, 2, 3, 4, 5, 6, 7, 8])

Get it:

Contribute:

Docs:

pip install pyrsistent

https://github.com/tobgu/pyrsistent

http://pyrsistent.readthedocs.org/