I'll go over it later today.

Any reason why the fix I've sent #13 isn't good?

After going through it quickly (not proving anything):

• I think one of the branches in the insert_from shifts existing elements to the right, creates a „hole“ in the middle and fills it with the items from the iterator. If the iterator is too short, even being careful to update the length after the fact instead of before won't solve it, since the actual elements are not consecutive in the array.
• It's not strictly for that CVE/Rustsec, but at least when I was writing Fix soundness issues in sized chunks and ringbuffer #13, I must have thought the ringbuffer has a very similar set of problems as the sized chunk. I'm not claiming to be sure to be correct in that feeling, but do you want to check that one too?

I haven't noticed #13 before! That PR looks great, so I've merged that one instead. Thank you for your help.