DEV Community

Cover image for  What is the ideal length of a code for you?

What is the ideal length of a code for you?

Mina Arroyo
I'm just a woman who loves to code.
・1 min read

In your work environment, does the length of code per function or per class matters? If yes, can you explain and give me some advice why does is important to follow. For example, the ideal length of the code you will create should be around 300 or 400 lines only.

Please leave some comment. I would love to hear your inputs. Thanks!

Photo by Rohit Farmer on Unsplash

Discussion (5)

alekseiberezkin profile image
Aleksei Berezkin

I heard somewhere about “magic number 7” — it's considered that a human may keep in mind no more than 7 things simultaneously. Translating to programming, any piece of code at any level should work with no more than 7 concepts. This may be number of locals, methods, fields, lines in one method or block etc.

kovah profile image
Kevin Woblick

I think it's hard to find a fixed number, but in general: if you have to scroll too much, you should split up the code. Functions itself should never reach a length where you have to scroll through the code (except for like 12" laptops maybe). It's not that hard to move structures like loops out of a function and move it to new function.

mattother profile image

It really depends who you talk to. I think it's more useful to ask what are the pro/cons for functions of varying length.

The biggest issue with longer functions is the scope of variable lifetimes gets bigger and bigger, so it becomes easier to couple variables in unintuititive ways. Ex.

def do_stuff():
    x = calculate_a_thing()

    # a bunch of lines of code...

    # has x changed?
    # is it still being used in the same context it was originally?
    # did somebody accidently start using it, not knowing it was already used ealier?
    # etc...
    f = another_process(x, y)

It's similar to the issues with global scope, it becomes harder to see where things are starting to couple in dangerous ways and usually becomes more difficult to maintain.

Also, the cognitive load required to retain everything that going on in the functions gets more and more difficult the longer it gets. I think if you look at a function and you're afraid to refactor it, that's probably a good sign that it's too long.

Some people like a consistent level of abstration too, personally I'm not convinced this is super important, but it can help to be cognizant of. Ex.

def build_something(values):
    summed = sum_the_values(values)
    process_the_values(summed, values)

# vs.

def build_something(values):
    sum = 0
    for v in values:
        sum += v
    process_the_values(summed, values)

Personally I do find readability suffers when methods are too small, but I like the idea of an IDE tool like xtofl pointed out.

Also, not directly related, but the number of arguments is usually a good sign a function is doing too much (and usually that an incorrect abstration is being used).

For classes, again really depends on the class, but if there are a lot of methods it's probably worth asking yourself if it's doing too much. I think better indicators are a large number of dependencies or functions lacking cohesion. Ex.

class DoingALot(object):

  def __init__(self, invoice_service, payment_gateway, datetime_service, token_factory, potatoe_generator):
    super(DoingALot, self).__init__()
    # ...

  def print_invoices(invoice_printer):
    # ...

  # If you're printing invoices and processing payments in the same class, your probably doing too much.
  def process_payment(payment_vo):
    # ...
xtofl profile image

I like it terribly short (under 10 lines); preferrably zero (most stuff has been written already).

Why? Because I find myself more often reading and trying to reverse-engineer a function's intent than trying to find all the details of every possible network interaction. Longer function bodies often have much repetition, nesting etc... which clutter the overview.

I found many people find the opposite: they want to see the details, and don't want to jump back and forth to find them.

But really, shouldn't we start having IDE's that make this irrelevant? Refactoring tools can extract details from long function bodies; I guess the code viewing tools could easily do the opposite: inline the details when you're from 'the wrong camp' ;).