John Stilley

The unit tests broke after my PR merged. Sorry!

Okay, we just had a meeting about this ticket. A big take-away from the meeting is (in the example above) we want to write the "0,3" time step along with...

> We would benefit from a standardized thing that sets `Reactor.p.time` That's an interesting point.

The reason for that is if you call `loadTestReactor()` and it has already been pickled, loading the test reactor is 20-50 times faster. It can really speed up unit tests....

Sigh... Arrielle has isolated the following bad situation downstream: 1. Someone builds the test reactor, and pickles it 2. The [global nuclides](https://github.com/terrapower/armi/issues/473) get tweaked by some test 3. The test...

> We could also cache a hash/checksum of the nuclide info and then check it upon unpack and invalidate/rebuild the cache if the hashes don't match. I could implement this...

I expect this ticket to be closed when this one is: https://github.com/terrapower/armi/issues/473

Order of Operations: 1. PR 1373 goes through. 2. Modify `HexGrid` to be two subclasses: Tips Up or Flats Up (or add an attribute in the class, or something) 3....

I think there may be a little confusion here. The distances defined by `x` and `y` here are the "flat-to-flat" distance, whether the `HexGrid` is flat-to-flat or "corners up": https://github.com/terrapower/armi/blob/6163e942abe873129533b84a6f2efa632f260b47/armi/reactor/grids/hexagonal.py#L110-L111...

I explicitly removed this. ARMI does not contain any "requirements", and thus the sphinx-needs tooling fails. As ARMI _can't_ move the requirements into this repo, it will never be possible...