Remove unused function testCase generating warning when compiling test modules.

[rbasso]

Some exercises show a warning when compiling the test module:

Defined but not used: `testCase`

This can confuse users and is not considered good practice.

Unused code:

• function testCase
• import of Assertion

Affected exercises:

• atbash-cipher
• binary
• crypto-square
• hexadecimal
• largest-series-product
• nth-prime
• prime-factors
• raindrops
• roman-numerals
• scrabble-score
• triangle

Closes #145

[kytrinyx]

I can't think of a reason to keep the code that (a) we are not using and (b) causes a warning.

@towynlin Is there something here that I'm missing?

[petertseng]

Now you are getting me curious as to why Travis doesn't fail without this PR, as we do run tests with -Wall -Werror...

Experimenting quickly on my machine, some raw results:

• neither runghc -Wall -Werror not runhaskell -Wall -Werror emit errors
• ghc -Wall -Werror fails as you would expect: binary_test.hs:11:1: Warning: Defined but not used: ‘testCase
• there are cases in which runhaskell will emit errors: if I have an unused binding in Binary.hs, only then we get something: Binary.hs:7:1: Warning: Defined but not used: ‘a’ - perhaps runhaskell does something to ignore unused errors from the main module... or something? I would benefit greatly from an explanation here!

As for this PR:

The pattern we will see is that we'll have two lines mentioning testCase in the file.

$ git grep testCase | cut -d: -f1 | uniq -c | sort -n
      2 atbash-cipher/atbash-cipher_test.hs
      2 binary/binary_test.hs
      2 crypto-square/crypto-square_test.hs
      2 hexadecimal/hexadecimal_test.hs
      2 largest-series-product/largest-series-product_test.hs
      2 nth-prime/nth-prime_test.hs
      2 prime-factors/prime-factors_test.hs
      2 raindrops/raindrops_test.hs
      2 roman-numerals/roman-numerals_test.hs
      2 scrabble-score/scrabble-score_test.hs
      2 triangle/triangle_test.hs
      3 sgf-parsing/sgf-parsing_test.hs
...

(further entries omitted, we are only interested in entries with 2 lines, that means we also don't touch sgf-parsing)

this seems to match the list in this PR, so this all seems good - it is certain that testCase is unused in the files mentioned in this PR (and if it were being used, Travis would have failed).

[petertseng]

I imagine the reason why testCase was present was... copypasta, probably.

The only test files that don't have testCase in them...

$ git grep -L testCase */*_test.hs
food-chain/food-chain_test.hs
house/house_test.hs
octal/octal_test.hs
trinary/trinary_test.hs

So probably some copy paste that includes testCase by default, even though it's not used everywhere (since it was used in a majority).

[rbasso]

Same result here regarding runhaskell. It shows some warnings but not others, but I don't have an explanation for that.

I got the warnings because I have a very different setup here. I don't like the use of a global installation of GHC with the packages provided by _test/bootstrap.sh, so I'm using stack projects for all exercises in my repository of solutions - even in Travis CI - to have better isolation, replicability and track dependencies more precisely.

I really prefer stack test over _test/check-exercises.

[petertseng]

To me personally, it seems valuable to track dependencies per exercise instead of globally. I definitely took that approach for a different project with multiple self-contained problems. I wonder if one day that would be an appropriate thing to do for this repo.

I realize what I said is not really relevant to this PR so maybe suited for a different issue.

[towynlin]

Thanks for pinging me @kytrinyx — haven't found the time to review yet. Good discussion here. Keep it up folks. I'm not hearing a conclusive merge/close call from the discussion yet. If anybody is confident one way or the other, please say so!

[petertseng]

For what it's worth, I would support merging this.

[rbasso]

We where just discussing others issues in the wrong place. Sorry!
About this PR I believe everyone here agrees.

[kytrinyx]

Cool, in that case... :merge:

[towynlin]

I see — finally got a minute to glance over the code. Appreciate your git | cut | uniq | sort data mining @petertseng — that makes very clear that this is the right change. Thanks @rbasso!