What is the rational behind this commit? I would expect a u8 because you can't rotate negative.

[hellow554]

What is the rational behind this commit? I would expect a u8 because you can't rotate negative.

_Originally posted by @hellow554 in 1aedb76

CC @ZapAnton

[ZapAnton]

The main rationale for the commit is this comment made by @coriolinus.

The purpose was, at least for me at the time, to give a student a chance to learn about the difference between signed/unsigned types and how to build / handle different APIs, built using these types.

Guess the above reasoning may sound too far-fetched, so perhaps if you discuss the input type with the maintainers, it could be changed back to u8

[hellow554]

Thanks for the explanation. Yes, I would argue, that a u8 would fit way better here. It's a bit weird to write c as u8 as i8 😄 .

[coriolinus]

I would expect a u8 because you can't rotate negative.

Why can't you rotate negative? That just means left rotation, instead of right rotation.

[hellow554]

Okay, let me correct the word can't to won't. If you want to keep i8, then I would suggest adding a testcase that tests exactly that.

[coriolinus]

A PR adding a testcase with negative rotation would be welcomed.

[ZapAnton]

Perhaps it would make sense to wait until exercism/problem-specifications#1438 is resolved and then add test cases to this track

[ZapAnton]

exercism/problem-specifications#1438 has been closed without merging, which means we are free too implement additional negative shift test cases ourselfs, without syncing with the problem-specifications repository. I will try to make a PR in the near future.

There is one point of concern though. The linked PR was closed, because the maintainers believed that these particular test cases would make the exercise more difficult and frustrating for the students. If we are too add the tests, should we mark them in some special way, so students may skip them, if they want to (feature-locked tests, comments in the code, section in the README, etc.)?

@coriolinus may I ask for you opinion on the above moment?

[coriolinus]

I don't believe it is necessary to feature-lock these tests. In my opinion negative numbers here are not an advanced concept.

If mentors discover a trend otherwise, we can revisit this issue in the future.