fix: display decimals on largest unit

[HHK1]

Fixes https://github.com/EvanHahn/HumanizeDuration.js/issues/150

As explained in the issue, using the library with largest set (with or without units specified), results in leaving the decimal part for units larger than second.

• This fixes this behaviour and adds unit tests
• I changed the way rounding was working since it leads to inaccurate results
• Added some tests regarding: https://github.com/EvanHahn/HumanizeDuration.js/issues/141 since it was linked to the issue above. I don't see any problems there though.

Now, some tests are breaking, specifically:

const h = humanizer({ largest: 2 })
assert.strictEqual(h(540360012), '6 days, 6 hours')
assert.strictEqual(h(540360012, { largest: 3 }), '6 days, 6 hours, 6 minutes')

I haven't touched those to avoid hiding this. I think it's reasonable to have those tests failed if round or maxDecimalPoints is not provided, the only issue is that it's going to be a "breaking" change in those cases.

You can leave it as is (and fix the test ofc), and publish a major version, or only apply the remainder if maxDecimalPoints has been provided, what do you think ?

[EvanHahn]

Thanks for this.

I won't have a chance to look at this for a few days, but I've written down to take a look soon.

[EvanHahn]

Thanks for doing this. I think this makes sense to do as a breaking change, so would you mind fixing the broken tests?

[HHK1]

Fixed the tests in https://github.com/EvanHahn/HumanizeDuration.js/pull/171/commits/53f05719470029d9b1be3fee097758c02b45fd61

I also realised that passing Number.Infinity was now causing a crash, so I added some tests and fixed it. Behaviour is the same as before, and is now tested: https://github.com/EvanHahn/HumanizeDuration.js/pull/171/commits/8433326dc8a74461306a6eab5979fa3df2633867

[EvanHahn]

Thanks. I plan to take another look in the next week or so.

[HHK1]

Hey @EvanHahn , any luck of having this part of the next release ? Is there anything I can do to help you review this ?

[EvanHahn]

Sorry that I've been slow to respond here. I'm trying to take a more holistic look at the library to address some bugs like this (among others).

[HHK1]

Hi @EvanHahn, just checking if you want me to update this, maybe target your v4.0.0 branch directly instead of master ?

[EvanHahn]

Sorry for moving slowly here. I'll get back to you.

[EvanHahn]

I just renamed master to main which caused this PR to be automatically closed. I didn't realize this would happen. Apologies!