Shouldn't this public function check for null?

[alas]

Shouldn't this public function check for null?

https://github.com/google/s2geometry/blob/ad5ef6c8cf4b0286868e3116e7347e2e114b4ff2/src/s2/s2lax_loop_shape.h#L76

context: https://github.com/google/s2geometry/blob/ad5ef6c8cf4b0286868e3116e7347e2e114b4ff2/src/s2/s2lax_loop_shape.h#L96

[jmr]

I don't think so. Could you explain why you think it should check for null?

The (undocumented, but hopefully apparent) precondition is 0 <= i && i < num_vertices(). Do you have an example test case that would be affected by your proposed change?

[alas]

I only asked cause it looks like a public facing API, I'm used to checking those for erroneous/malicious user input.

[smcallis]

This is a pretty common pattern in C++, in general you don't want to pay the cost for bounds checking. We have the new iteration API that prevents this is in the common case anyways. If we did bounds check then we have the question of what we would return when it fails. We don't use exceptions so we'd have to return a std::optional or absl::StatusOr probably, which would be a breaking change.

[alas]

closing then, ty