Composition

Overview

The ease by which other people can read and understand a program (often called "readability" in software engineering) is among the most important qualities of a program. Readable programs are used and extended by others, sometimes for decades. On this matter, the authors Hal Abelsen and Jerry Sussman of Structure and Interpretation of Computer Programs wrote, "Programs must be written for people to read, and only incidentally for machines to execute."

Excellent composition does not mean adhering strictly to prescribed style conventions. There are many ways to program well, just as there are many styles of effective communication. However, the following guiding principles universally lead to better composition of programs:

  • Names. To a computer, names are arbitrary symbols: "xegyawebpi" and "foo" are just as meaningful as "tally" and "denominator". To humans, comprehensible names aid immensely in comprehending programs. Choose names for your functions and values that indicate their use, purpose, and meaning. See the lecture notes section on choosing names for more suggestions.
  • Functions. Functions are our primary mechanism for abstraction, and so each function should ideally have a single job that can be used throughout a program. When given the choice between calling a function or copying and pasting its body, strive to call the function and maintain abstraction in your program. See the lecture notes section on composing functions for more suggestions.
  • Purpose. Each line of code in a program should have a purpose. Statements should be removed if they no longer have any effect (perhaps because they were useful for a previous version of the program, but are no longer needed). Large blocks of unused code, even when turned into comments, are confusing to readers. Feel free to keep your old implementations in a separate file for your own use, but don't turn them in as your finished product.
  • Brevity. An idea expressed in four lines of code is often clearer than the same idea expressed in forty. You do not need to try to minimize the length of your program, but look for opportunities to reduce the size of your program substantially by reusing functions you have already defined.

Names and variables

Variable and function names should be self-descriptive:

Good

goal, score, opp_score = 100, 0, 0
greeting = 'hello world'
is_even = lambda x: x % 2

Bad

a, b, m = 100, 0, 0
thing = 'hello world'
stuff = lambda x: x % 2

Indices and mathematical symbols

Using one-letter names and abbreviations is okay for indices, mathematical symbols, or if it is obvious what the variables are referring to.

Good

i = 0         # a counter for a loop
x, y = 0, 0   # x and y coordinates
p, q = 5, 17  # mathematical names in the context of the question

In general, i, j, and k are the most common indices used.

'o' and 'l'

Do not use the letters 'o' and 'l' by themselves as names:

Bad

o = O + 4     # letter 'O' or number 0?
l = l + 5     # letter 'l' or number 1?

Unnecessary variables

Don't create unnecessary variables. For example,

Good

return answer(argument)

Bad

result = answer(argument)
return result

However, if it is unclear what your code is referring to, or if the expression is too long, you should create a variable:

Good

divisible_by_49 = lambda x: x % 49 == 0
score = (total + 1) // 7
do_something(divisible_by_49, score)

Bad

do_something(lambda x: x % 49 == 0, (total + 1) // 7)

Profanity

Don't leave profanity in your code. Even if you're really frustrated.

Bad

eff_this_class = 666

Naming convention

Use lower_case_and_underscores for variables and functions:

Good

total_score = 0
final_score = 1

def mean_strategy(score, opp):
    ...

Bad

TotalScore = 0
finalScore = 1

def Mean_Strategy(score, opp):
    ...

On the other hand, use CamelCase for classes:

Good

class ExampleClass:
    ...

Bad

class example_class:
    ...

Spacing and Indentation

Whitespace style might seem superfluous, but using whitespace in certain places (and omitting it in others) will often make it easier to read code. In addition, since Python code depends on whitespace (e.g. indentation), it requires some extra attention.

Spaces vs. tabs

Use spaces, not tabs for indentation. Our starter code always uses 4 spaces instead of tabs. If you use both spaces and tabs, Python will raise an IndentationError.

Many text editors, including VS Code and Atom, offer a setting to automatically use spaces instead of tabs.

Indent size

Use 4 spaces to denote an indent. Technically, Python allows you to use any number of spaces as long as you are consistent across an indentation level. The conventional style is to use 4 spaces.

Line Length

Keep lines under 80 characters long. Other conventions use 70 or 72 characters, but 80 is usually the upper limit. 80 characters is not a hard limit, but exercise good judgement! Long lines might be a sign that the logic is too much to fit on one line!

Double-spacing

Don't double-space code. That is, do not insert a blank line in between lines of code. It increases the amount of scrolling needed and goes against the style of the rest of the code we provide.

One exception to this rule is that there should be space between two functions or classes.

Spaces with operators

Use spaces between + and -. Depending on how illegible expressions get, you can use your own judgement for *, /, and ** (as long as it's easy to read at a glance, it's fine).

Good

x = a + b*c*(a**2) / c - 4

Bad

x=a+b*c*(a**2)/c-4

Spacing lists

When using tuples, lists, or function operands, leave one space after each comma ,:

Good

tup = (x, x/2, x/3, x/4)

Bad

tup = (x,x/2,x/3,x/4)

Line wrapping

If a line gets too long, use parentheses to continue onto the next line:

Good

def func(a, b, c, d, e, f,
         g, h, i):
    # body

tup = (1, 2, 3, 4, 5,
       6, 7, 8)
names = ('alice',
         'bob',
         'eve')

Notice that the subsequent lines line up with the start of the sequence. It can also be good practice to add an indent to imply expression continuation; use whichever format expresses the line continuation most clearly.

Good

total = (this_is(a, very, lengthy) + line + of_code
            + so_it - should(be, separated)
            + onto(multiple, lines))

Blank lines

Leave a blank line between the end of a function or class and the next line:

Good

def example():
    return 'stuff'

x = example() # notice the space above

Trailing whitespace

Don't leave whitespace at the end of a line.

Control Structures

Boolean comparisons

Don't compare a boolean variable to True or False:

Bad

if pred == True:   # bad!
    ...
if pred == False:  # bad!
    ...

Instead, do this:

Good

if pred:           # good!
    ...
if not pred:       # good!
    ...

Use the "implicit" False value when possible. Examples include empty containers like [], (), {}, set().

Good

if s:         # if s is not empty
    ...
if not tup:   # if tup is empty
    ...

Checking None

Use is and is not for None, not == and !=.

Redundant if/else

Don't do this:

Bad

if pred:            # bad!
    return True
else:
    return False

Instead, do this:

Good

return pred         # good!

Likewise:

Bad

if num != 49:
    total += example(4, 5, True)
else:
    total += example(4, 5, False)

In the example above, the only thing that changes between the conditionals is the boolean at the end. Instead, do this:

Good

total += example(4, 5, num != 49)

In addition, don't include the same code in both the if and the else clause of a conditional:

Bad

if pred:            # bad!
    print('stuff')
    x += 1
    return x
else:
    x += 1
    return x

Instead, pull the line(s) out of the conditional:

Good

if pred:            # good!
    print('stuff')
x += 1
return x

while vs. if

Don't use a while loop when you should use an if:

Bad

while pred:
    x += 1
    return x

Instead, use an if:

Good

if pred:
    x += 1
    return x

Parentheses

Don't use parentheses with conditional statements:

Bad

if (x == 4):
    ...
elif (x == 5):
    ...
while (x < 10):
    ...

Parentheses are not necessary in Python conditionals (they are in other languages though).

Comments

Recall that Python comments begin with the # sign. Keep in mind that the triple-quotes are technically strings, not comments. Comments can be helpful for explaining ambiguous code, but there are some guidelines for when to use them.

Docstrings

Put docstrings only at the top of functions. Docstrings are denoted by triple-quotes at the beginning of a function or class:

Good

def average(fn, samples):
    """Calls a 0-argument function SAMPLES times, and takes
    the average of the outcome.
    """

You should not put docstrings in the middle of the function -- only put them at the beginning.

Remove commented-out code

Remove commented-out code from final version. You can comment lines out when you are debugging but make sure your final submission is free of commented-out code. This makes it easier for readers to identify relevant portions of code.

Unnecessary comments

Don't write unnecessary comments. For example, the following is bad:

Bad

def example(y):
    x += 1            # increments x by 1
    return square(x)  # returns the square of x

Your actual code should be self-documenting -- try to make it as obvious as possible what you are doing without resorting to comments. Only use comments if something is not obvious or needs to be explicitly emphasized.

Repetition

In general, don't repeat yourself (DRY). It wastes space and can be computationally inefficient. It can also make the code less readable.

Do not repeat complex expressions:

Bad

if a + b - 3 * h / 2 % 47 == 4:
    total += a + b - 3 * h / 2 % 47
    return total

Instead, store the expression in a variable:

Good

turn_score = a + b - 3 * h / 2 % 47
if turn_score == 4:
    total += turn_score
    return total

Don't repeat computationally-heavy function calls either:

Bad

if takes_one_minute_to_run(x) != ():
    first = takes_one_minute_to_run(x)[0]
    second = takes_one_minute_to_run(x)[1]
    third = takes_one_minute_to_run(x)[2]

Instead, store the expression in a variable:

Good

result = takes_one_minute_to_run(x)
if result != ():
    first = result[0]
    second = result[1]
    third = result[2]

Semicolons

Do not use semicolons. Python statements don't need to end with semicolons.