Notes on programing languages

Interpreters on 2023

program - parse -> abstract-syntax-tree - eval -> result

sspy.py - a tiny scheme interpreter with python - shows the details.

Python Data Model -> A Framework/API for core language constructs

We can leverage the Python Data Model to build new classes.

By using and implementing special methods of Python Data Model in your objects, your objects can behave like the built-in types, enabling the expressive coding style Pythonic. You will find that your intuition applies everywhere.

Data Sturctures

• sequences - Python sequences are often categorized as mutalbe or immutable, and also could be considered as flat sequences and container sequences.

  • flat sequences like str, bytes, array.array store the value of its contents in its own memory space

    • more compact because of the physically store the value

  • container sequences like list, tuple, and collections.deque

    • hold references to the objects

  • list -> mutable and mixed-type

    • list comprehensions and generator expression
    • given a list t, list(t) must create a new copy of t

  • tuples as immutable lists

    • tuple(t) just returns a reference to the same t, there's no copy here
    • a tuple's length will never change, and a tuple uses less memory than a list

  • array.array -> efficient because of only the packed bytes for numeric data

    • for large sequences of numbers, this saves a lot of memory
    • NumPy lib

• collections

  • dict and set

    • hash tables are the engines for the high performance dicts
    • hash tables must be sparse to work, they are not space efficient, compared to a low-level array a pointers to its elements (more compact but also much lower to search)

  • Data class as a collection of fields

• str versus bytes

• Variables are mere labels

  • the parameters in the function are aliases of the actual arguments

Functions as objects

• Functions, like integers, strings, and dictionaries, also can be a program entity, this enables programming in a functional style.
  • The main idea is to assign functions to variables, pass them to other functions, store them in data structures and access function attributes.
• function decorators and closures
  • a decorator like deco is a function with the decorated function parameter func

target = deco(target)

def deco(func):
    def inner():
        print("running inner()")
    return inner

@deco
def target():
    print("running target()")
    return

Object Oriented

• duck typing
  • it's a sequence because it implements the necessary sequence methods

Iteration

• iterator

  • iterables have a iter() method that builds a new iterator each time.
  • iteators implement a next() method that returns individual items

• Generators are iterators that produce the values passed to yield

  • yield -> return the generator object

    • yield pauses the function and saves the local state so that it can be resumed right where it left off

  • a generator function gen_ab builds a generator object gen_ab() that implements the iterator interface

def gen_ab():
    yield 'A'
    yield 'B'

• Concurrency

Context Managers

• context manager objects exist to control a with statement

  • The context manager interface consists of the enter and exit methods

  • using @contextmanager decorator instead of creating a class and implementing the interface enter/exit methods

generators as coroutines

Metaprogramming

Standard ML

• functional programming

• sml-ch1

  • syntax is how you write sth

  • semantics is what that something means

    • Type-checking rules (before program runs) in current static environment
      • what type a binding has
      • produces a type or fail

  • Evaluating the bindings (in the dynamic environment)

    • a value or an error or an infinite loop of the preceding bindings
    • look up value in current dynamic environment

  • idioms are the common approaches to using langguage features

    • Recursion
    • Let -> local binding

  • libraries

    • standard

  • tools

    • REPL -> quick try-something-out
    • debugger

  • immutation data

    • it is just a mapping, not assignment statement, a tuple, or a list
    • No constructs for mutating the data we have build. No way to change the contents of a binding, a tuple,or a list
    • don't worry about the alias or copy like in java
    • or, like java, you have to care whether alias or copy, and in order to avoid the mutable data is been changed

• sml-ch2

  • tuples are syntactic sugar for records with field names 1, 2, ...

  • type synonyms -> a convenience for talking about types

  • datatypes bindings

    • patter-matching over one-of types
    • better to use pattern-matching to access list and option

  • patter matching over each-of types

    • value binding -> a val-binding can use a pattern, not just a variable

    • function binding

      • in ML, every function takes exactly one tuple arg, implemented with a tuple pattern in the function binding
      • a function argument can also be a pattern

    • Type-checker can figure out the types of things you're matching against, no longer need to write down any explicit types for the arguments to functions or any variables

    • type inference

      • more general can replace its type variables consistently

  • nested pattern-matching

    • elegantly recursive

  • Recursion

    • tail-recursive -> recursive calls are tail-calls
    • call-stacks -> pop the caller before the call, allowing callee to reuse the same stack space

• sml-ch3

  • first-class functions
  • TODO

notes

• Reload modules problem in Emacs Python Shell

  • use importlib to reload(models)
  • or use ipython and %autoreload

• virtualenvs setup for python3 -> pipenv

  • New a project: >pipenv --python 3.6
  • Install all dependencies: >pipenv install
  • Locate the virtualenv: >pipenv --venv
  • Use the shell: >pipenv shell
  • Uninstall everything: >pipenv uninstall --all

• Your environment contains PYTHONPATH=/usr/local/lib/python2.7/site-packages This doesn't work with Python 3 for obvious reasons. To remove it: unset PYTHONPATH

reference

• How to Design Programs, 2nd
• Fluent Python, 2nd by Luciano Ramalho