The Elements of Style in Ruby #5: Readability of long numeric literals

Today’s topic is the following rule from the Ruby Style Guide:

Add underscores to large numeric literals to improve their readability.

Most of the programs we write feature a substantial number of numeric literals(e.g. 10, 0.34, 0b1010, 0123, 0xCAFE). There is nothing strange or unusual about that. From time to time, however, those literals are pretty long:

MAX_SIZE = 10000000000
DEVIATION = 0.2343434343
BIT_MASK = 0b100101010101

I’m pretty sure most of you would have pretty hard time to quickly digest a number written in this way - lots of digits and no separators between them to help us discern the number’s structure. At this point _ makes a dramatic appearance and comes to the rescue:

MAX_SIZE = 10_000_000_000
DEVIATION = 0.2_343_434_343
BIT_MASK = 0b1001_0101_0101

The addition of a few _ improves the readability of those huge literals a ton!

The underscores we add to numeric literals are ignored by Ruby:

100_000
# => 100000

As you can see from the preceding example nothing’s lost or changed - we’ve only gained readability and eased the parsing burden on our brains.

Obviously we should not overdo _:

# short literals are pretty readable on their own
# bad
1_00

# good
100

Personally, when dealing with decimal literals, I tend to use _ for numbers with at 5 least digits (e.g. 11_948). The number of digits to separate with _ depends on the numeric base - in decimal it makes sense to group digits by 3 (e.g. 1_000_000), in binary by 4 (e.g. 0b1111_1010_1110), etc.

That’s all for today folks! Hope I managed to convince at least a few of you of the benefits of using underscores in your long numeric literals.

As usual I’m looking forward to hearing your thoughts here and on Twitter!