What is the purpose of this test?

[drewj-tp]

I stumbled ontest_gridNumPinsAndLocations and it kind of tripped my brain.

armi/armi/reactor/tests/test_blocks.py

Lines 2439 to 2464 in 661a11d

	def test_gridNumPinsAndLocations(self):
	b = blocks.HexBlock("fuel", height=10.0)
	fuelDims = {"Tinput": 25.0, "Thot": 600, "od": 0.76, "id": 0.00, "mult": 168.0}
	cladDims = {"Tinput": 25.0, "Thot": 450, "od": 0.80, "id": 0.77, "mult": 168.0}
	ductDims = {"Tinput": 25.0, "Thot": 400, "op": 16, "ip": 15.3, "mult": 1.0}
	wireDims = {
	"Tinput": 25.0,
	"Thot": 600,
	"od": 0.1,
	"id": 0.0,
	"axialPitch": 30.0,
	"helixDiameter": 0.9,
	"mult": 168.0,
	}
	wire = components.Helix("wire", "HT9", **wireDims)
	fuel = components.Circle("fuel", "UZr", **fuelDims)
	clad = components.Circle("clad", "HT9", **cladDims)
	duct = components.Hexagon("duct", "HT9", **ductDims)
	b.add(fuel)
	b.add(clad)
	b.add(duct)
	b.add(wire)
	with self.assertRaises(ValueError):
	b.autoCreateSpatialGrids(self.r.core.spatialGrid)
	self.assertIsNone(b.spatialGrid)

It's testing that creating the grid should fail, but it isn't immediately clear from the test why it should fail.

The magic number 168 is the problem. Someone deeply versed in magic hexagon numbers will know that 169 is a tiling number for hexagons: you can fit 169 closely packed circles into a hexagonal lattice and fully fill seven rings.

And the test does fail because 168 actual pins (components with mult == 168) does not match the expected full lattice pins of 169.

And the name of the test implies something with locations but there's nothing really with locations asserted?

If the point of the test is to check autoCreateSpatialGrids refuses to make an incomplete hex lattice, I think we should do the following:

1. Use a number that is much further away from a magic hex tiling number (e.g., 150)
2. Change the test name to be test_autoGridFailsWithPartialGrid or something
3. Add some more commentary (doc string, raise regex string, etc.)

I'm not making the PR now because I'm not sure if there's intended to be more in the test. I'm happy to if my understanding of what the test is supposed to do is correct.

[john-science]

A poorly documented test. My apologies. I didn't write the first version of it, but I've touched it twice.

You can't create a spatial grid if you mix items with multiplicity 1 with items of multiplicity N.

From the auto create spatial Friday docstring:

Raises
        ------
        ValueError
            If the multiplicities of the block are not only 1 or N or if generated ringNumber leads
            to more positions than necessary.