Repository navigation
I think the Matrix problem would be better without the 'string' tests. #541
Description
Activity
Even though the https://github.com/exercism/x-common/blob/master/exercises/matrix/canonical-data.json does not exist yet, what @derifatives says is true - the https://github.com/exercism/x-common/blob/master/exercises/matrix/description.md only says matrices of numbers, so the https://github.com/exercism/xhaskell/blob/master/exercises/matrix/test/Tests.hs#L76-L84 seem out-of-place.
If the maintainers of this track decide that the parsing questions are useful to have, they can be moved to a track-specific exercise, perhaps?
Reacted by Ivan FedyuninAnd my personal opinion:
Will have to wait until I've done this exercise.
@derifatives wrote:
Basically, I think the type inference is coming from the other side of the shouldBe in the tests, is that right?
I don't understand what you mean by "from the other side", but here is how I understand it:
let intMatrix = fromString :: String -> Matrix Int
intMatrixis defined as a specialized version of the used-definedfromString, having typeString -> Matrix Int.it "extract first row" $ do row 0 (intMatrix "1 2\n10 20") `shouldBe` vector [1, 2]
Knowing the type of
intMatrixandrow, the compiler can directly deduce that the expression on the left side ofshouldBeis of typeVector Int, which is the same type as the right side.I also don't feel it makes much sense in the context of the problem.
Yeah! It feels weird there!
It seems that
fromStringis there since the original commit that added the exercise.I agree that there are reasons to remove it:
- It isn't needed for the exercise to make sense.
- It doesn't seem to add anything interesting to the exercise.
- Exercises should probably be focused in a few concepts.
- There seems to be no sensible implementation that isn't a partial function.
The only problem is that it isn't possible to remove
fromStringwithout rewriting the entire test suite, which may be a good idea...Did the exercise, doing the string cases took more time than the entire rest of the problem, I think. Then again, I used the lowest-effort solution possible for the rest of the problem.
readsskills not strong enough, apparently. It was an OK thing to think about, but good for its own exercise.Reacted by Ivan Fedyunin
As a beginner-intermediate Haskeller making my way through the exercises, this was the first time I had to submit an incomplete result and look up the answer. When I looked at it I didn't understand it, and I'm still not sure I understand it. [Basically, I think the type inference is coming from the other side of the
shouldBein the tests, is that right?] I never would have figured this out for myself.I also don't feel it makes much sense in the context of the problem. The direct text of the problem is "Given a string representing a matrix of numbers, return the rows and columns of that matrix." Expecting that to work with a matrix of strings is not reasonable, particularly given that it's a much more difficult parsing problem than anything seen so far.