state-of-tic-tac-toe: unique descriptions for test cases

[jiegillet]

Follow-up to #2012.

It was fairly easy to come up with natural descriptions, except for the draws, so I used numbers for those two cases.

[SaschaMann]

This doesn't add any info nor does it explain why the case is there, I'd remove the numbering.

Uniqueness is not guaranteed or required in any way.

[ErikSchierboom]

Uniqueness is not guaranteed or required in any way.

I would strongly argue against having duplicate descriptions. Yes, they are (currently) not required, but I'd like them to be (I think almost all canonical data adheres to this).

As for this particular case, maybe have the first description be "Draw" and the second one "Another draw"?

[jiegillet]

I'm fine with changing the description to "another draw", and I would also push towards having unique descriptions, it really helps with generators, or in case of reimplementation to see which test needs to be changed, etc.

[SaschaMann]

That's redundant. It's obvious that in a test group that contains draws, as stated by the group description, all tests will be a draw or another draw. Unless there's an explicit reason for why a certain case exists, e.g. an edge case of a draw, this is useless info and needlessly clutters up the file when scanning and reading it.

[SaschaMann]

(You could make an argument that there's no reason to have two drawn games that don't cover any particular edge cases in the first place but that ship has sailed now.)

[jiegillet]

Well yes, a redundant naming convention for a redundant situation. The test case is fine BTW, maybe it will be crucial to a student who had an off-by-one error or something.

I don't really mind either way, I just want something unique so it's easier for people down the line.

[SaschaMann]

see above

[petertseng]

I feel strongly that diagonals should be referred to as falling and rising rather than first and second, because there is no clear ordering to the two diagonals relative to each other. Those changes are highly suggested.

I don't feel strongly about the rows and columns but I think it'd be clearer to refer to them by their position rather than implying an ordering. But I will understand if the consensus decision is that first, second, and third is better for rows and columns. This is because I understand that there is an unambiguous ordering for rows and columns since the language Exercism uses for its written materials is English, which has a clearly-defined order (top-to-bottom and left-to-right).

[SaschaMann]

I feel strongly that diagonals should be referred to as falling and rising rather than first and second, because there is no clear ordering to the two diagonals relative to each other. Those changes are highly suggested.

The mathematical terms would be main diagonal and antidiagonal fwiw: https://en.wikipedia.org/wiki/Main_diagonal. I'm not sure if rising and falling has any precedence. Though I'd be fine with either.

[jiegillet]

Thank you @petertseng, it really is better this way.

I left falling and rising, because it's more descriptive than the proper mathematical description.

[ErikSchierboom]

Thanks!