Add 'piecing-it-together' exercise #2554
Conversation
|
Solving this was surprisingly difficult. It did uncover two errors in the data, though. |
|
Yes, it is one of those exercises that only seem simple. I would assume a difficulty level of 6-7, depending on the language. |
There was a problem hiding this comment.
@atk I haven't been following the discussion over the past few days. Is there a specific reason why there are only five test cases? I expected more tbh.
|
How many test cases would you deem sufficient? |
The exact number doesn't really matter, but I expect the current tests to be reasonably sufficient and cover most cases. It's a little difficult to assess whether that's the case. If you think these tests are enough, I'm fine with it — we can always add more in the future. |
There was a problem hiding this comment.
I just noticed that the aspect ratio is a float in all tests, which introduces floating-point operations. Would it be better to restrict it to integers (and let tracks add extra tests if they want to), or do we intentionally want floating-point numbers in this exercise? Maintainers?
Edit: asking this because the new test case has aspect ratio 2/3
|
Uneven aspectRatios are pretty normal in reality. Only square puzzles or 8-pieces baby puzzles (4/2) usually come with an integer aspect ratio. As an alternative, we could use a rational number stored in an 2-element array, but then it would need to be reduced to the lowest terms in order to avoid ambiguity, which would then result in an overlap with the rational-numbers exercise - at least that was my reasoning to choose a float here. |
|
I don't disagree. The question is whether floating-point manipulation is something we actually want in this exercise. It's realistic, yes, but the exercise may choose not to deal with this scenario for various reasons — making the exercise easier being one of them — while still allowing tracks to introduce extra tests with floats, if they choose to. |
The new test uses a truncated float, but this only results in an approximation of the columns rather than the actual integer that appears in the result. So it seems that a rational number is the only option here — or perhaps there's another solution I'm not aware of. |
|
I used the output of a calculation in node.js, which should be the closest to what JSON is supposed to handle in terms of f64. I don't think using a rational number for the aspect ratio is going to simplify the exercise significantly in most languages. Yes, you do not have to handle IEEE-754 numbers and their imprecisions then, but instead you need to reduce to the lowest term, which is already covered by another exercise. |
|
Using a truncated float means that you will arrive at an approximation of an integer. Rounding should work here. |
It does. I think something like this should be mentioned in a comment or similar. Let's wait until other maintainers chime in, since I'm not sure what the best way to handle this is. |
|
I think floating point values are fine to have in the test data. |
|
I think we can leave the choice to the maintainers. I have updated the documentation accordingly. |
|
@ErikSchierboom Can you please review when you get the chance? |
|
This is no show stopper but something to think about: maybe use a whimsical name? Puzzling Puzzles or something like that. Someone else might be able to come up with something better. Other than that, and once the images are included, this is great |
Co-authored-by: Erik Schierboom <erik_schierboom@hotmail.com>
Co-authored-by: Anastasios Chatzialexiou <16361161+tasxatzial@users.noreply.github.com>
Co-authored-by: Anastasios Chatzialexiou <16361161+tasxatzial@users.noreply.github.com>
Co-authored-by: Anastasios Chatzialexiou <16361161+tasxatzial@users.noreply.github.com>
Co-authored-by: Anastasios Chatzialexiou <16361161+tasxatzial@users.noreply.github.com>
Co-authored-by: Anastasios Chatzialexiou <16361161+tasxatzial@users.noreply.github.com>
Co-authored-by: Erik Schierboom <erik_schierboom@hotmail.com>
Co-authored-by: Erik Schierboom <erik_schierboom@hotmail.com>
|
I think it may make it fractionally more engaging to be less abstract with "complete" and say "discover" here:
English is confusing. To say "If it contains contradictions" may be misinterpreted as having to contain multiple contradictions. Adding
€0.02++ |
|
I changed it a bit to make it less ambiguous while still matching the theme. |
After starting a new jigsaw puzzle last weekend, I had the idea for a new simple exercise: a helper to provide information about jigsaw puzzles from a few pieces of information.
This is the first problem-specifiation I created, so I assume there are a lot of mistakes in there.