DEV Community

3 Common Mistakes that Python Newbies Make

Ryan Palo on August 04, 2018

Last weekend, I stared mentoring people on exercism.io on the Python track. I wasn't sure what to expect, but over the last week I have mentored a...

Read full post

Shreyas Minocha • Aug 5 '18

I think that

if year % 400 == 0:
    return True
if year % 100 == 0:
    return False
if year % 4 == 0:
    return True
return False

... is much clearer than

if year % 4 == 0 and (year % 100 != 0 or year % 400 == 0):
    return True
else:
    return False

... or even

return (year % 4 == 0 and (year % 100 != 0 or year % 400 == 0))

Of course this wouldn't always be true but I think that we should make these decisions on the basis of code readability and clarity of intent rather than always going by the number of SLOC or the length of lines.

Ryan Palo • Aug 5 '18

I personally agree. This example is probably one Boolean too long for a one-liner. I just wanted to show that technique. For some reason, most people I mentor can understand the one-liner better. I don’t know why. Maybe it’s easier for them to understand than inverting the order of comparisons in their head.

Valentin Berlier • Aug 5 '18 • Edited

Using the expression and putting it in its own function is usually pretty clear. It lets you give a label to the condition. You can also put each comparison on its own line if the expression is too long.

def is_leap_year(year):
    return (
        year % 4 == 0
        and (
            year % 100 != 0 
            or year % 400 == 0
        )
    )

The logical operators stand out more when they're at the beginning of a line.

Miguel Ruiz • Sep 7 '18

I agree too

Corey McCarty • Jan 11 '19

Raymond Hettinger said "one thought per line" and I think this is the best guide for direction with things like this. You can have several things going on as long as they can be thought of (in the realm of your code) as a single thought.

Thomas Landin • Aug 5 '18 • Edited

One thing you see all the time from people with previous programming experience is trying to make "C-style" for loops iterating over indices and subscripting the list, i.e. for i in range(len(list)) instead of the idiomatic for item in list.

Thomas Landin • Aug 5 '18

Oh, and generally working against the language and its idioms by shoehorning things into place instead of picking an appropriate data structure or language construct from the outset.

Examples:

Wanting dicts to be ordered instead of using OrderedDict (this is changed in Python 3.7+).

Looping through dicts and accessing elements by key instead of using the appropriate iterator.

Using for loops and indexing to access parts of lists instead of using slices.

Things like that.

Idan Arye • Aug 5 '18

dict ordering is important because keyword arguments are stored in a dict. Consider this:

$ python3.4
Python 3.4.8 (default, Jun 23 2018, 02:35:33) 
[GCC 8.1.1 20180531] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from collections import OrderedDict
>>> OrderedDict(a=1, b=2, c=3)
OrderedDict([('a', 1), ('c', 3), ('b', 2)])
>>> OrderedDict(c=1, b=2, a=3)
OrderedDict([('a', 3), ('c', 1), ('b', 2)])

We created an OrderedDict, but the constructor was using **kwargs which means the arguments we passed to it were stored in a regular dict and got reordered before OrderedDict.__init__ could store them in order.

With Python 3.6+, the **kwargs stores the arguments in the same order they were written, and the OrderedDict construction works as expected:

$ python3.6
Python 3.6.6 (default, Jun 27 2018, 13:11:40) 
[GCC 8.1.1 20180531] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from collections import OrderedDict
>>> OrderedDict(a=1, b=2, c=3)
OrderedDict([('a', 1), ('b', 2), ('c', 3)])
>>> OrderedDict(c=1, b=2, a=3)
OrderedDict([('c', 1), ('b', 2), ('a', 3)])

rhymes • Aug 5 '18

A little detail on ordered dicts:

Technically dicts have been ordered since 3.6 on (C)Python, it's just that in 3.7 they decided that to be call "Python 3.7" any implementation has to have the insertion order.

So, dicts in Python 3.6+ are ordered only in the official version, which means that you might have portability issues if you depend on that behavior, but in 3.7+ you are garanteed they will be ordered, regardless of the implementation.

Difference with Python 2:

Python 2.7.15 (default, Jul  3 2018, 10:08:31)
[GCC 4.2.1 Compatible Apple LLVM 9.1.0 (clang-902.0.39.2)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> a = {'b': 1, 'c': 3, 'a': '4'}
>>> a
{'a': '4', 'c': 3, 'b': 1}
>>>

Python 3.6.6 (default, Jul 20 2018, 18:34:07)
[GCC 4.2.1 Compatible Apple LLVM 9.1.0 (clang-902.0.39.2)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> a = {'b': 1, 'c': 3, 'a': '4'}
>>> a
{'b': 1, 'c': 3, 'a': '4'}

Thomas Landin • Aug 5 '18

Yep, that's what I meant by 3.7 being different on that point.

Fabio Rosado • Aug 5 '18

I see a lot of this when people want to use the index to do something instead of using for i in enumerate(foo) which returns (index, value)

Thomas Landin • Aug 5 '18

enumerate is a great tool and it's what you should use when you actually need the index for something, but most of the time people don't really need the index.

And there's a better way to use it:

for index, value in enumerate(foo):
    # voilà, the values are already unpacked
    # into two handy variables!

Vlastimil Pospichal • Aug 5 '18

This is a bit faster, because the first condition is met most often:

if year % 4 != 0:
    return False
if year % 100 != 0:
    return True
if year % 400 != 0:
    return False
return True

Shorter version:

if year % 4 != 0:
    return False
if year % 100 != 0:
    return True
return year % 400 == 0

Shortest version:

return year % 4 == 0 and (year % 100 != 0 or year % 400 == 0)

Ryan Palo • Aug 5 '18

That’s a good point! I feel like that order would be more intuitive for students as well! It does have the slight downside of using “negative comparisons” but I think the ease of understanding definitely outweighs that. 😁

Vlastimil Pospichal • Aug 5 '18

Negative comparisons are needed to eliminate else from source codes.

Ryan Palo • Aug 5 '18

I’m not 100% sure that else is a bad thing. But I agree that using negative comparisons isn’t a terrible thing. Maybe more of a “all things being equal, try to use positive rather than negative if possible” kind of thing.

Vlastimil Pospichal • Aug 5 '18

Eliminating else made my source code cleaner and faster. I use else in ternary operator only. I transformed structure if-else into ternary expression in many times.

Some languages (XSLT for example) are without "else". This inspired me.

Fabio Rosado • Aug 5 '18

I understand where you come from with your example to check if a year is a leap year. It's a good example since you do need to use quite a few comparisons to figure out if a given year is a leap year. I will have to disagree with you that using a one-liner is a much clearer or even less redundant.

The thing is, if you encounter a code with this style in a massive code base:

if year % 400 == 0:
    return True
if year % 100 == 0:
    return False
if year % 4 == 0:
    return True
return False

With your eyes, you can figure out what is happening there in less than a second. But when you encounter a one-liner like this:

return (year % 4 == 0 and (year % 100 != 0 or year % 400 == 0))

You actually need to stop and in your head, you have to disassemble the logic on that statement. The same thing happens with generators, if you are creating a massive generator that makes the code less readable is better to just use the long form.

One thing worth mention is that when you do code challenges it is quite fun to try and scale down the solution to a one-liner - some platforms even reward you for coming up with a short solution, so it's good to know how to scale down a long if statement like that.

Ryan Palo • Aug 5 '18

Yes, this is very true. I mentioned in the article that the one-line Boolean is probably getting a little long and dense. Personally, I really like the multiple return version a lot better too. I show the Boolean version to illustrate that as one option for removing nesting. Good point about doing shortest solutions as a challenge :)

λlvaro Frias • Aug 8 '18

Hey, Great post! Reading it i realized that my most recurrent mistake is try to use lists for everything. Checking out your examples of sets and dicts i see their potential as alternatives data structures. Thank you!

Ryan Palo • Aug 8 '18

Glad I could help 😁 thanks for reading!

Jason C. McDonald • Aug 5 '18 • Edited

You forgot my 'favorite'!

sudo pip install ...

Perhaps the best book I've ever read on this subject is How to Make Mistakes in Python by Mike Pirnat, a free eBook from O'Reilly [not to be confused with O'Rly ;) ]

Ben Sinclair • Aug 6 '18

General case of using sudo for every command.

Ryan Palo • Aug 5 '18

Ah! I haven’t read that one! Thank you!! It’s on the list. 😃

Alexandr K • Aug 5 '18

not sure, how I missed set syntax in Python :)

Thanks!

Esteban Rocha • Aug 6 '18

Good article and all... but that image it's too way disturbing, I mean it was really necessary to put a disgusting image of an animal doing an aberrant act that's been documented as a symptom of captivity? this is a Dev blog, seriously why the disgusting pic at all?

Ryan Palo • Aug 6 '18

Interesting, I had no idea this was a symptom of distress. Here's some info for anyone who is confused.

I was going for a combination of Python (the language) and eating your own tail (as in shooting yourself in the foot, getting yourself into trouble, causing problems for yourself, etc.). Obviously, if I knew it was a sign of animal distress, I wouldn't have included it. Thanks for the friendly heads up! I'll get that changed.

Martin Häusler • Sep 2 '18

I'm not a Python programmer, but in general I find return statements which produce a boolean value increadibly hard to read, unless the returned value is a constant (return true / return false) or an aptly named variable. I personally try to avoid return statements where the right side contains any boolean operators.

Ankur Tiwari • Jan 18 '20

Useful!

Ryan Palo • Jan 18 '20

Thanks! Glad it helped!