QuantityDimension: Fixed divide(Dimension dim) where dim is a product

[DrDaleks]

That is basically a follow up on #124, where the following tests would fail:

Dimension m2 = QuantityDimension.MASS.pow(2);
Dimension result = QuantityDimension.LENGTH.divide(m2);
assertEquals("[L]/[M]²", result.toString()); // <= returns "[L]/null" instead

Dimension ml = QuantityDimension.MASS.multiply(QuantityDimension.LENGTH);
Dimension result = QuantityDimension.LENGTH.divide(ml);
assertEquals("1/[M]", result.toString()); // <= returns "[L]/null" instead

This is due to divide(Dimension dim) calling multiply(dim.pow(-1)) which has changed behaviour after #129

---

This change is 

[DrDaleks]

having second (deeper) thoughts. Putting that on hold for now.

[DrDaleks]

So here's the thing. #129 changed the handling of powered units in a way that any unit (and therefore pseudounit) calling a pow(int n) encapsulates the original unit, thereby a) fixing formatting issues in SimpleUnitFormat (see #124), but b) preventing the canonisation process.

The calls to ProductUnit.ofQuotient(a, b) circumvent the issue, but the question is: is this the right path, or should we revert #129 and have a profound refactoring of SimpleUnitFormat instead (which I don't see as simple in it's current state).

I'd rather be in favor of pushing this further (for now) and returning to this issue (if it is one) at a later time

[keilw]

Are the tests working now?

[DrDaleks]

Yes, I've added 2 unit tests for the issue (both passing).

[DrDaleks]

I did however found other (similar) issues related to unit.pow(-1) outside of the current (dimension) scope. Will probably make another PR soon for those cases.

[keilw]

@DrDaleks Thanks, also for considering to join the JSR. It seems, you are not a JCP Member yet. Are you employed by your company (it is not a JCP member either) or work as a freelancer/contractor? Permanent employees usually need their employer to sign the JSPA and agreements to join a JSR as Expert, if that was not easy or possible, please join as Associate Member. I approved it but depending on which member category you get, it could be necessary to re-apply as Contributor (unless the PMO could tweak that but I'm not sure)