Is your hope that the presence of the error: will trigger some CIs to fail, causing people to update sooner?

I think we need to ponder the wording carefully, as well as the timing of requiring --force but the concept is good. I don't personally have a problem with the wording you chose, but I'm aware that in the past my judgement on these kinds of things isn't great. I wonder if we could get some help from the compiler's error messages team?

The error won't cause a exit code of 1, but it will pollute stderr. I hope folk will notice it and update. I intend to file a bug on cross and get them to update immediately, so that they are ready for our next release where we conceptually do the actual enforcement.

A lot of people shouted at me that I'd break CI systems for emitting a deprecation error (but not failing to run) because of systems scanning stderr for 'error:' so I guess they'd notice pretty quickly if that's the case. Honestly this is such a corner case of use, providing cross works properly that I don't mind.

I wonder if we should have a specific flag for this though, since --force implies more than just this particular override. Again I don't mind if you think we should keep things simple.

Talking to people about the error message sounds like a fine idea; perhaps we can ask at the wg-sync? I can add a -not-host suffix to this flag easily, and its nice future proofing, so sure lets do that.