Inconsistencies with associativity of multiplication.

[dex-k]

Hi, I just came across Numbat from this post, and went to try it out. After skimming the documentation, I tried to do a currency conversion:

>>> 1 usd * ( aud / usd )

  1 dollar × (australian_dollar / dollar)

      = 1 AUD    [Money]

>>> aud / usd

  australian_dollar / dollar

      = 0.665332

>>> 1 usd * 0.665332

  1 dollar × 0.665332

      = 0.665332 $    [Money]

>>> 1 * usd * (aud/usd)

  1 dollar × (australian_dollar / dollar)

      = 1 AUD    [Money]

>>> 1 * (aud/usd) * usd

  1 × (australian_dollar / dollar) × dollar

      = 0.665332 $    [Money]

Having read the (seemingly) relevant parts of the documentation again, I don't understand why all of the former snippets don't evaluate to the same result. Is this intended behaviour?

[Goju-Ryu]

This looks a bit odd. I think it might be associated with the issues discussed following this comment on another issue.

And just in case you are unaware, it looks like you are trying to convert between currencies which is much more easily done this way: 1 usd -> aud or with the keyword to instead of the arrow if you prefer.

[dex-k]

Thank you, that's helpful to know.

[sharkdp]

What is the exact issue here?

>>> 1 * usd * (aud/usd)

  1 dollar × (australian_dollar / dollar)

      = 1 AUD    [Money]

>>> 1 * (aud/usd) * usd

  1 × (australian_dollar / dollar) × dollar

      = 0.665332 $    [Money]

These two results do not look the same, but they are equal, as you can confirm by running:

>>> 1 * (aud/usd) * usd == 1 * usd * (aud/usd)

    = true    [Bool]

Yes, the unit may change when changing the order or multiplication (do you mean "commutativity" instead of "associativity"?). And we should probably change that. For example, we could always choose the "smaller" unit. Or we could always choose the "more fundamental" unit (and if impossible, fall back to some internal ordering of units). Is this what's bothering you?