There is need to translate existing Objective-C code: on platforms where the Objective-C bridge is unavailable (Linux and Windows), Swift code can not call Objective-C code, nor toll-free bridge with Objective-C types, and therefore Objective-C types cannot be subclassed.

For non-Apple platforms, you have to use a different implementation of Foundation — which has pretty good but not identical coverage to Apple's one.

As a side note, it's always seemed clear to me that when Apple provides a new technology, they provide a stop-gap solution, like Carbon, for developers who, for a variety of possible reasons, may not be able or willing to use the new tech, like Cocoa.

The bridge between Objective-C and Swift has always seemed like Carbon to me.

Fortunately, unlike when Carbon was canned and rewriting for Cocoa seemed like a real chore, I don't think Apple will can the Objective-C bridge any time soon — but for Swift code now targeting or intending to target non-Apple platforms, it may be best to start thinking about migrating away from relying on the bridge now — decarbonify your Swift code.

Doesn't seem like there would be much demand to run legacy Objective-C code on other platforms, as it would most likely be strongly tied into macOS or iOS APIs that would be availble anyway.

Indeed. Although, some of the things that Apple's ObjC Foundation provides that aren't in Swift's cross-platform Foundation are sorely missed — but I doubt there's much impetus to put them in Swift's cross-platform Foundation, either.

Yeah, that could be an interesting use case.