04 January 2019 0 comments Javascript

**tl;dr; Use Number.prototype.toString() to display percentages that might be floating point numbers.**

I started writing a complicated solution but as I discovered corner cases and surprised I was brutally forced to do some research and actually read some documentation. Turns out `Number.prototype.toString()`

, with the `precision`

argument omitted, is the ideal solution.

The application I was working on has an input field to type in a percentage. I.e. a number between 0 and 100. But whatever the user types in, we store the number in decimal. So, if the user typed in "10" into the input widget, we actually store it as `0.1`

in the database. Most people will type in a whole number (aka. an integer) like "12" or "5" but some people actually need more precision so they might type in "0.2%" which means `0.002`

stored in the backend database.

But the widget is a React *controlled* component meaning it's `value`

prop needs to be potentially formatted to what gives the best user experience. If the user types in whole numbers set the `value`

prop to a whole number. If the user types in floating point numbers set the `value`

prop type a floating point number with the "matching formatting".

I started writing an overly complicated function that tries to figure out how many decimal-points the user typed in. For example `0.123`

is 3 because `parseInt(0.123 * 10 ** 3, 10) === 0.123 * 10 ** 3`

. But, that approach doesn't work because of floating point arithmetic and the rounding problem. For example `103441 !== 10.3441 * (10 ** 4) === 103440.99999999999`

. So, don't look for a number to pass into `.toFixed()`

.

Turns out ```
Number.prototype.toString()
```

is all you need. If you omit the precision argument, it figures out how many significant digits to use based on the input. It's best explained with some examples:

> (33).toString() "33" > (33.3).toString() "33.3" > (33.10000).toString() "33.1" > (10.3441).toString() "10.3441"

Perfect!

So actually, it's a bit more complicated than that. You see, the number stored in the backend database might be `0.007`

which you and I know as "0.7%" but be warned:

> 0.008 * 100 0.8 > 0.007 * 100 0.7000000000000001

You know, because of floating-point arithmetic, which every high-level software engineer remembers understanding one time years ago but now know just to watch out for.

So if you use the `toString()`

on that you'd get...

> var backendPercentage = 0.007 > (100 * backendPercentage).toString() + '%' "0.700000000000001%"

Ouch! So how to solve that? Use `Math.round(number * 100) / 100`

to get rid of those rounding errors. Apparently, it's very fast too. So, now combine this with the `toString()`

:

> var backendPercentage = 0.007 > (Math.round(100 * backendPercentage * 100) / 100).toString() + '%' "0.7%"

Perfect!

- Previous:
- Concurrent download with hashin without --update-all 18 December 2018
- Next:
- Use vars() to send an argparse Namespace into a function in Python 08 January 2019