Repost detection fails when URL is not exactly the same as was originally submitted, so should consider broadening the matching a bit.
E.g. When the original article is submitted as https://example.com/
and someone tries to submit https://example.com
they receive no repost warning. So at the very least /
should probably be ignored as part of the detection. However some other common examples of this happening are with Youtube vs Youtu.be links, and other shorteners, as well as AMP vs non-amp links.
Edited by cfabbro