Tests combinations max-min fractionDigits and Significant digits

[romulocintra]

#3508 (comment)

Todo's for other tests files:

• Combine "notation" : "compact" with all other values
• validate resolvedOptions for each set of options

The goal's to have correct matrix results aligned with spec

cc @sffc @FrankYFTang

[ptomato]

I think the test matrix approach looks good, I trust that the champions know best what the correct format outputs should be 😄 Currently they are all the same though, I assume that's because it's a draft and the real values will be entered later?

One improvement might be to use keys that explain what that particular line tests. e.g.

var optionsMatrix = {
  'same-minimums': { useGrouping: false, minimumSignificantDigits: 2, minimumFractionDigits: 2 },
  'min-sd-larger': { useGrouping: false, minimumSignificantDigits: 3, minimumFractionDigits: 1 },
  ...

and

'auto': {
  'same-minimums': { '1': '1.0', ...

[ptomato]

I think this is basically correct — I didn't check the rounding results exhaustively, but I sampled maybe about 40% of them.

If you wanted to add some comments to explain what is going on in each entry, I'm sure that would really help implementors debug their implementations. For example, in the expected results matrix,

'morePrecision': {
  // 2-21 significant digits
  'same-minimums': { ... },
  // 1-2 significant digits or 0-2 fraction digits, whichever is more precise
  'same-maximums': { ... },
  // 3-21 significant digits
  'minSd-larger-minFd': { ... },

and in the options matrix, for example,

// max significant digits wins
'same-minimums': { ... },
// max digits equal, result with more precision wins
'same-maximums': { ... },

etc.

[ptomato]

What is supposed to happen in this case? My understanding is that maximumSignificantDigits and maximumFractionDigits are both 3, so it depends on the rounding result itself, which one wins the conflict. Is that correct?

[ptomato]

OK by me, I still think the comments I suggested would be useful if you have time to add them, but what you have is already fine.

[romulocintra]

OK by me, I still think the comments I suggested would be useful if you have time to add them, but what you have is already fine.

If it's ok for you I can work on those comments while doing pending tasks in upcoming PRs:

• Combine "notation" : "compact" with all other values
• validate resolvedOptions for each set of options

That way we can land current tests