Sync nucleotide-count with problem-specifications #550
Conversation
This removes an unimplemented test from the tests.toml.
This deletes some tests that are not in the canonical-data in problem-specifications.
This reorders the tests to match the order in the canonical-data.
This reformats the config.json to make configlet happy.
This updates the existing tests to use the inputs and test names from canonical-data.
This brings the nucleotide-count exercise up to date with the specification in problem-specifications. The sync brought in changes to the instructions and metadata, as well as adding a new test.
The tests no longer reference the count function. This deletes it.
kytrinyx
changed the title
|
I completely missed this back in November. Previously implementing this task with a class made sense. A Now that the member function To me that looks now like a task for a regular function that performs the validation (it might |
Probably not! I was likely just following what was already there, and then tweaking it to follow the current canonical data. I don't actually write C++, and I don't even remember making this PR (which is a bit disconcerting). Please feel free to change the implementation, if you have something better in mind. |
Before PR exercism#550 the constructor was required to validate the input, the member function `count(char)` had to return the count for a specific nucleotide, the member function `nucleotide_counts()` had to return a `std::map` with all four counts. But PR exercism#550 has changed that significantly. After syncing the tests with the canonical data there's no call to `count(char)` anymore, so each test looks like a two-step function call: ``` const nucleotide_count::counter dna("G"); const auto actual = dna.nucleotide_counts(); ``` I'd call that unnecessarily complicated and unidiomatic, there's no longer a good reason to require a `class`. Let's turn the class `counter` into a function: ``` const auto actual = nucleotide_count::count("G"); ```
Before PR #550 the constructor was required to validate the input, the member function `count(char)` had to return the count for a specific nucleotide, the member function `nucleotide_counts()` had to return a `std::map` with all four counts. But PR #550 has changed that significantly. After syncing the tests with the canonical data there's no call to `count(char)` anymore, so each test looks like a two-step function call: ``` const nucleotide_count::counter dna("G"); const auto actual = dna.nucleotide_counts(); ``` I'd call that unnecessarily complicated and unidiomatic, there's no longer a good reason to require a `class`. Let's turn the class `counter` into a function: ``` const auto actual = nucleotide_count::count("G"); ```
This brings the nucleotide-counts exercise up to date with problem-specifications.
For clarity, it makes the change in several commits:
tests.tomlthat are not reflected in the test suite (per Update each tracks tests.toml file #370)canonical-data.jsoncanonical-data.jsonconfiglet fmton nucleotide-count to normalize theconfig.jsonconfiglet syncfor the exercise and implementing the missing test.Closes #544
Closes #539