The Python Object Protocol

A "good" object protocol, allows one to control meta, hence being able to modify the behaviour and the notation of the language itself. Python achieves this through dunder methods. Allowing python to be a multi-paradign language

Make use of dunder methods for better readability, and extensibility

Special cases aren't special enough to break the rules.
Although practicality beats purity.

Providing an implementation of dunder len and getitem allows the class to be more explicit, in the sense it holds some data.

It is also worth mentioning a builtin's len method != dunder len, since directly reads the ob_size field in python bytecode

Infix methods, such as ones those enable LHS and RHS representations, such as dunder le, gt etc should also be provided if is appropriate to do so

Muggle Methods: a feature not generally publicized, but makes

Containers

• listcomps and gencomps are preferable over filter and map which largely are remnants of legacy python, it is almost always better to use explicit generators over such functions

• array.array is a thin-wrapper around C's continuous memory aka Arrays. Great for storing homogeneous data, at low memory, since a list needs to allocate extra space to amortize the cost of a append() call.

• Tuple type also saves on memory as it is immutable, and is stored as a fixed-sized array. A tuple (which doesn't hold mutable type) is known as runtime.

• Pythons match..case should be used for matching the structure of an object, arbitiary conditions.

Dictionaries

• ChainMap merges all dictionaries to form a singular look-up-table. Updating this, auto-updates all the objects held in the table
• hashable in python is anything that has a dunder hash function and doesn't change across the runtime

Data Classes

• collections.namedtuple is albeit a legacy feature that extends the base tuple class to add support for infix, repr, comparasion and other "good to have" operations on what was otherwise a bare class.

• typing.NamedTuple is a newer implementation and also supports type-checking through static code analysers

• use dataclasses as a scaffold for classes with the intention of adding custom methods later. Relying solely on dataclasses might lead to a code smell, especially if there are many "dumb" classes

• make use of slots when heavily contrained. It instead of allowing dict like class access, sets the elements at runtime and doesn't allow modifiication. Saving memory