Is Shorter Code, Better Code?

Depends. I think a good example is to look at the daily Leetcode for Dec 2. (1266 if you're interested). Also, spoiler alert!

Here's the more readable version that I would like to see in someone's code.

res = 0
for i in range(len(points) - 1, 0, -1):
       y_diff = abs(points[i][1] - points[i - 1][1])
       x_diff = abs(points[i][0] - points[i - 1][0])
       res += max(y_diff, x_diff)
return res

And here's a version that runs a little bit faster due to inbuilt functions, but is much less readable:

return sum(max(abs(x1-x2), abs(y1-y2)) for (x1, y1), (x2, y2) in zip(p, p[1:]))

So... you decide. If you come back to the second version in a year or two, are you going to be happy with yourself? Probably not. However, maybe an alternative that's even better than the first two options is to simply leave good comments, or just leave the readable version of the code, but commented.

From my beginner perspective, I'd say that depends on whether someone else has to read and understand it. The short version may be pretty unintelligible. Using descriptive variable names with intermediate steps, for example, could lengthen the code but also strengthen the understanding.

One thing I've noticed in myself is that it's very easy to intuitively think that shorter code does less and therefore performs better. And even when we should know better, writing a short one-liner in a scripting language sometimes just feels like a lot less is happening, than writing the equivalent code in a lower-level language.

To me this is specially apparent when I go back and forth between writing Ruby and Lua; where a snippet of code feels good in Ruby, because it's easy to express, but in Lua it feels terrible, because I often have to first implement stuff that Ruby does out of the box, so I get to actually see all the performance implications some code has.

Suddenly, calling a method can become looping over ancestor classes, doing a hash lookup for each of them and ultimately calling the subroutine.

As for readability, it's hard to say. People sometimes obsess over brevity, and I think it's not a bad thing in general. As you start shortening code, it often becomes more readable at first, until you reach a tipping point where shortening the code further ends up making it less readable again.

Generally speaking, if your code looks complicated, long and messy, it's almost always one of two cases:

1) Your code is more complex than the inherent complexity of the problem and you should probably work on simplifying it further, so it becomes clear what it does to someone who understands the task it's trying to solve.

2) The problem you are trying to solve is inherently very complex, and reducing the code further would just hide the complexity by spreading it around arbitrary abstractions that don't really help much. In this case, the best solution is usually to add comments explaining how the code reflects the solution.

Keep in mind though, that most online coding exercises will rarely have problems that fit into category 2, for a variety of reasons.

I'd say shorter as long as it's readable.

Simple example:

return empty($someVar) ? true : false;

can be simplified to

return empty($someVar);

=> shorter but still highly readable.

If you have a highly optimized piece of code inside a function with a good name then it's totally fine.

Example from "Perl Golf" of a short but difficult to read one-liner:

$a=1;print$a-=$b+=$a*=-1,$/for 0..31

It cannot be shorter but it cannot be less explicit either. Put it inside a printFibonacciNumbers function and it becomes clearer.

As long as it's not sacrificing readability, it's ok. However, readability can depend on the conventions of the language.

Perl is an interesting example. The language is very terse and if you're not familiar with the language it's hard to follow. However, once you understand the relevant parts, such as the use of the default variable and the reliance of regexes for string handling it makes perfect sense.

Is soup made from a soup-starter better soup than soup created from raw ingredients?

"better" is always relative.

If I want to impress my inlaws I will make my own soup.
If I need to feed 3 hungry children tonight, I will use a soup-starter

I recommend better code.
Shorter code is not important. The most important thing is scalable, readable, well-optimized.
Even though the code is short, if its performance is bad, it's just useless code.

it depends on the definition of 'shorter'. is it just lines of code? for instance, would you consider:

$j  = $j >= 1 ? 1 : 0;

to be 'shorter' than

if($j >= 1) {
    $j = 1;
}
else {
    $j = 0;
}

even though it's the exact same number of statements?

i think counting lines of code is always a dangerous metric. i can easily (and, , often do) write very terse code that is difficult for others to read. there are better measures of complexity, ie 'cyclomatic complexity' (which isn't great. but is still better than just counting lines)

blog.ndepend.com/understanding-cyc...

Better for what? Unless you specify that, the question is too ambiguous to allow for a useful answer.

For smaller bundle sizes? Definitely. For performance? Depends. For maintainability? There's a sweet spot before maximal conciseness, where you put at most three functionalities in one line, so they could be understood in a normal reading flow.

As a rule of thumb, if you need to comment the what and not the why to make the code understandable, you're beyond that sweet spot. Only go there if it is a required optimization.

From a real life example, I once wrote a lease tracking module for an accounting package right after suffering a horrible injury. I was taking a great deal of morphine at the time that I wrote the core of the calculation engine. Turns out it was a single line of code. It passed every test I threw at it, could calculate lease interest and payments out to fifty years to 5 decimal places.

To this very day, I still have no idea how the f*ck it works...