Clock: Does the test force a non-idiomatic solution?

[guygastineau]

Hi all,

While doing the clock exercise I noticed:

• The test expects #at to be a class method.
• In order to deal with tests of the form (Clock.at(12, 12) + 12).to_s the solution requires making an instance of the class.
  • This means most solutions call Clock.new inside def self.at.
  • I am not an expert, but this test seems like it forces a hack rather than an idiomatic solution.
• With a single class method I was able to pass all tests except those tests that added minutes after the class method call, so I am wondering if they should be excluded.
• Or perhaps the test should be refactored to work with a Clock class object instead of using class methods.
• I'm thinking that refactoring the test not to use class methods would make the solutions more idiomatic while allowing all of the test to be present.

Thank you

[guygastineau]

If this type of hack is idiomatic for Ruby then please feel free to tell me so.

Like I said above - it just seems like a hack to me.

[Insti]

Clock description Clock tests (Ruby) Clock canonical data

Hi @guygastineau, thanks for raising this issue. It's something that I've also been thinking about recently too.

As an experienced Ruby dev using a class method as an alternate constructor is not uncommon. Have a think about this: What is the difference between Clock.new and Clock.at?

That said I'm not a fan of positional arguments, and would rewrite it to use keyword arguments: Instead of Clock.at(23,59) I'd do Clock.new(hour: 23, minute: 59)

I also think the tests should be changed, I didn't like that minute addition was done with a Clock instance + a Fixnum, and would rather see it be Clock + Clock

Rather than:

  def test_add_minutes
    assert_equal "10:03", (Clock.at(10, 0) + 3).to_s
end

We'd have:

   def test_add_minutes
     assert_equal '10:03', (Clock.at(10, 0) + Clock.new(minutes: 3)).to_s
   end

The exercise description is pretty sparse, so I think we have some scope for rewriting the tests how we like.

I think it would be a good idea to rewrite the clock tests.

I'm happy to help you if you want to take a shot at this. What do you think?

P.S. - It would also be good (but not 100% necessary) to write a generator for the tests while we're at it.

[guygastineau]

@Insti thank you.

Yeah, I might have some time to work on this this week.

I'm also busy getting the bash tests.sh up to date with the generator I built for it. It is simpler in structure than the generator framework for Ruby and the templates are just ERB files

That being said. I would definitely like to make a generator for this test.

1. I will work on a generator for affine-cipher first, as I wrote that one. I think that will be good experience for me before taking on clock.
2. Your understanding of idiomatic Ruby will likely be required to make the clock test more natural.
   • I like your suggestions for it.
   • Is there more you would like to add to those suggestions?
3. I am more than willing to make the generator for clock based off of your insights into the clock exercise.

[kotp]

A student told me recently that the Clock exercise reminds them more of a Timestamp than a clock.

[guygastineau]

@kotp I can see that.

When I first started this exercise I saw the to_s's in the test and thought it was the built-in #to_s.

If you all agree I think that should be renamed to be clearer and more descriptive. - also to something without an already reserved namespace.

[Insti]

I would definitely like to make a generator for this test.

Great! :heart: Let me know if you have any questions along the way. Feel free to make a WIP (Work In Progress) Pull Request once you have some code to discuss.

[guygastineau]

How would you all feel about these two thing:

1. Instead of Clock.at(12, 12).to_s using Clock.at(12, 12).format ? # Of course the arguments could still be keyword arguments instead of positional.
2. Instead of (Clock.at(10, 0) + 3).to_s using (Clock.at(hour: 10, minute: 0) + Clock.at(minute: 3).minutes).format
   • I suppose this way kind of suggests the students implement attr_reader to access the total minutes from another clock object which could allow for future tests in which the added Clock object also has been given hours, eg.
   • (Clock.at(hour: 10, minute: 0) + Clock.at(hour: 12, minute: 43).minutes).format
   • I suppose in this case the def +(minutes) would still function the same as it does now, but there would be some extra functionality overall from the other changes. Some extra clarity too I think.

Let me know what you all think.

[emcoding]

Sounds nice! One more consideration in favor of keeping to_s, from a curriculum overview point of view: AFAIK the Clock exercise is the first exercise in the track that explicitly requires to implement a custom to_s. And I think that is very useful to get experience with. (I always forget about those kind of things.)

In general I like where this is going, but I don't know the Clock exercise very well.

[guygastineau]

@F3PiX

Thank you for your input.

So, it sounds like you are saying that it is idiomatic to implement a custom to_s method inside a class for Ruby?

I am familiar with duck punching, although it seems discouraged. I recognized that this is not duck punching.

I suppose to me it seems more sane not to mess with the namespace reserved for to_s, but if doing so would be idiomatic then I really don't have the experience to argue about it.

[guygastineau]

I will submit a WIP PR soon. I would like to get a few things more or less decide first:

• [x] Clock.new will replace Clock.at
• [x] A custom to_s will be kept
• [x] Adding minutes in the test will use another clock object, eg. (Clock.new(hour: 12, minute: 12) + Clock.new(minute: 12)).to_s

[Insti]

0. Get rid of Clock.at use Clock.new(hour: 1, minute: 2)

1. Instead of Clock.at(12, 12).to_s using Clock.at(12, 12).format ? # Of course the arguments could still be keyword arguments instead of positional.

Definitely use .to_s. It is idiomatic ruby.

2. Instead of (Clock.at(10, 0) + 3).to_s using (Clock.at(hour: 10, minute: 0) + Clock.at(minute: 3).minutes).format

Lord no!

Add Clock objects directly.

(Clock.new(hour: 10, minute: 0) + Clock.new(minute: 3)).to_s

I suppose this way kind of suggests the students implement attr_reader to access the total minutes from another clock object which could allow for future tests ....

This is implementation detail that we should not be testing. There are several ways to implement a solution, some of which won't do this. The student is free to write their own intermediate tests if they want.

[guygastineau]

@Insti

Thank you for the feedback. I guess I was confused about how the other object would be added to first one.

Since it is passed as a parameter to the first Clock obj I guess the + method can access it's methods to retrieve the information however the student wants.

Thanks for getting me to think about that. I think I have it figured out.

UPDATE: I edited the checklist above. Let me know if that looks good now.