grade-school: align with specification per #503

[wolf99]

Fixes #503

[wolf99]

/format

[ryanplusplus]

This test is really confusing. How does this ensure that a student can't be in two different grades? "Aimee" is added to grades 1 and 2, but then it checks that grade 5 is empty.

[wolf99]

Whoops, copy-pasta error

[wolf99]

Looking at this some more, the way this test is specified makes me think I should rearchitect this exercise completely.

My first inclination here was to check student names as a student is added, compare it to names already on the list and not add it if the name already exists.

However this supposes a linear timeline of added the input (i.e. add Aimee at grade 2 that succeeds, then attempt to add Aimee at grade 1 that fails). However, the desired grade is specified as 2 with an expected output of zero (implying that actually the addition of Aimee in grade 2 should have failed).

This suggests a much simpler interface like the following, where there is no local store of the roster, it is instead supplied as an input each time a query is made.

typedef struct {
   uint8_t grade;
   char *name;
} student_t;

typedef struct {
   size_t count;
   student_t students[MAX_STUDENTS];
} roster_t;

roster_t get_roster(roster_r students);

roster_t get_grade(uint8_t grade, roster_r students);

[wolf99]

I realised my idea for simplification of the interface (#555 (comment)) conflicts with the information in the exercise README.md.
I think the current interface is a bit ugly, making the tests more convoluted than they need to be; but I don't see another way to meet both the readme and the canonical-data.

So I have added a check for multiple names in the same grade in the current implementation in the latest commit instead.
Now the exercise and the implementation should match the problem specification and the readme.

[wolf99]

This still isn't right 🙃
Opening an issue on problem specifications as the spec is too conflicting

exercism/problem-specifications#1763

[wolf99]

/format

[wolf99]

Finally! This exercise is at a state it is possible to implement and match the spec 🙌
Ready for review.. Configlet is failing but this is for the icons for which there is #675 .