> Literally means "something violates the safety model".
Yes. But the safety model is not capable of identifying whether something is or is not safe in every case, just that it can't prove that it is safe. There are plenty of things you must do in unsafe that are safe, just not provably by the compiler. Thus, needing to use unsafe does not always imply actual unsafe operations.
If someone was under the impression that they were recommended to use unsafe but in a safe way but there were other caveats to the usage they weren't aware of, bad things could happen. I don't believe it's sufficient in a guide to rely on the fact that unsafe is recommended to convey the level of danger that recommendation entails.
Yes. But the safety model is not capable of identifying whether something is or is not safe in every case, just that it can't prove that it is safe. There are plenty of things you must do in unsafe that are safe, just not provably by the compiler. Thus, needing to use unsafe does not always imply actual unsafe operations.
If someone was under the impression that they were recommended to use unsafe but in a safe way but there were other caveats to the usage they weren't aware of, bad things could happen. I don't believe it's sufficient in a guide to rely on the fact that unsafe is recommended to convey the level of danger that recommendation entails.