Default largestUnit to the largest non-zero unit

[ryzokuken]

From #978 (comment),

I like this idea of "top-heavy". Would it be correct to say that if largestUnit is missing, it defaults to the largest nonzero unit in the duration? If so, then would we be able to apply a classic, deterministic balance algorithm (convert everything to the smallest unit, then balance up to the largest unit)?

/cc @sffc @justingrant

[justingrant]

Would it be correct to say that if largestUnit is missing, it defaults to the largest nonzero unit in the duration?

More or less. Here's the planned behavior from #856:

7. We'll add an 'auto' value for largestUnit in Duration and non-Duration difference methods where it's used. 'auto', like undefined, will select the default unit. EDIT 2020-10-06: added this section

• 6.1 On Duration methods, 'auto' means "the largest nonzero unit in the input(s)"

• 6.2 On difference of non-Duration types 'auto' means to pick the default unit of the type:

  • 6.2.1 Instant: 'seconds'
  • 6.2.2 Date / DateTime - 'days'
  • 6.2.3 ZonedDateTime - 'hours'
  • 6.2.4 Time - 'hours'
  • 6.2.5 YearMonth - 'years'

• 6.3 If the default chosen above is smaller than the user-provided smallestUnit then set largestUnit = smallestUnit.

@ryzokuken, does this proposed behavior above match what you had in mind?

If so, then would we be able to apply a classic, deterministic balance algorithm (convert everything to the smallest unit, then balance up to the largest unit)?

Not sure. @ptomato is closer to this problem as he's implementing the changes agreed in #856 (comment), so he'll likely have a more informed opinion. One caveat might be that there's some special-casing around weeks (see #856 (comment)) that probably means that the inputs to a "classic, deterministic balance algorithm" might require a boolean inputHasWeeks parameter to know whether the output should include weeks or not.

[ptomato]

What needs to be done in this issue?

[sffc]

Mainly docs I think, and verify that the spec and polyfill uses this behavior.

[ptomato]

I'm not 100% sure what "this behavior" is — does it apply only to Duration methods with a largestUnit parameter?

[justingrant]

I'm not 100% sure what "this behavior" is — does it apply only to Duration methods with a largestUnit parameter?

AFAIK this is correct: round, add, subtract are AFAIK the only methods affected.

AFAIK, this issue is part of #856 and can be closed, unless it's better to leave this open because #856 includes a bunch of other stuff too. The balancing.md page has already been updated for #856 behavior, and docs for round, add, subtract should be updated when their #856 code changes are made.

[ptomato]

In #1071 I've made a separate abstract operation out of the logic described here, and it's used in Duration's round, add, and subtract methods, so I'll close this.