Add custom nth_back to Skip

[acrrd]

Implementation of nth_back for Skip.
Part of #54054

[rust-highfive]

r? @cramertj

(rust_highfive has picked a reviewer for you, use r? to override)

[cramertj]

r? @scottmcm

[timvermeulen]

Looks correct to me, though I don't think we want nth_back to skip any of the elements at the front, similar to how next_back doesn't skip any. I think we can call self.iter.nth_back(n) as long as n < self.len(), without having to involve self.n.

[acrrd]

Didn't notice that, I changed it

[scottmcm]

Look good, thanks!

@bors r+ rollup

[bors]

📌 Commit 1edecacf2021a3381f447c949e6dc3018f4b0be4 has been approved by scottmcm

[bors]

⌛ Testing commit 1edecacf2021a3381f447c949e6dc3018f4b0be4 with merge 8e9bb42c1edf098a01f7a18737f23ad257978854...

[scottmcm]

@bors r-

Good point, @sinkuu; Thanks!

@acrrd Can you add some more tests for bigger n, and maybe some with by_ref to ensure behavior doesn't change here?

[kennytm]

@bors r- retry

[acrrd]

@sinkuu thanks!

@scottmcm I added the test for a bigger n and another test that check that nth_back behave as a non skipped iterator. It's not clear to me how to do the tests you meant using by_ref

[timvermeulen]

I don't think we want nth_back to skip any of the elements at the front, similar to how next_back doesn't skip any

Isn't this still a concern with the current implementation?

[acrrd]

@timvermeulen thanks :)

[sinkuu]

In case n >= self.len() && self.len() != 0, additionally you need to do self.iter.nth_back(self.len() - 1); (or whatever that takes self.len()-items from the back) since the underlying iterator could have side effects.

letmut v = vec![];letmut it = [1,2,3,4,5].iter().cloned().inspect(|x| v.push(*x)).skip(1);assert_eq!(it.len(),4);assert_eq!(it.nth_back(4),None);assert_eq!(v, vec![5,4,3,2]);

[acrrd]

Why let _ = and not just the call to nth_back?

[sinkuu]

It was just for expressing intent to throw away the return value of nth_back, and I don't meant it is actually needed.

[scottmcm]

I don't think the comment above was addressed; if disagree please add a test that demonstrates that the underlying iterator is still advanced.

(You could do that with inspect, or something like let mut it = ...; it.by_ref().skip(10).nth_back(100); assert_eq!(it.next_back(), ...);)

[acrrd]

Ok, I didn't understand you meant that.

[timvermeulen]

You're now calling self.len() three times, maybe it's better to store it in a variable first.

[jonas-schievink]

Ping from triage, @acrrd what's the status of this? Are there still open review comments to be addressed, or is this waiting on someone to approve?

[acrrd]

@jonas-schievink I think all comments are addressed, waiting for approval.

[scottmcm]

Sorry for the delay re-reviewing, this looks great now!

@bors r+ rollup

[bors]

📌 Commit e80a375 has been approved by scottmcm