> It's identical to R7RS, fortunately; the exception is that the object
> you raise must be of type &condition (and preferably of subtype &error
> if it will always be raised non-continuably).  Not all R6RS systems
> enforce this requirement.

We could just specify that R6RS systems add a &foreign condition type.

However, the situation is complicated by the fact that not all foreign
statuses are errors. Hence it would be misleading to make &foreign a
subtype or &error. We could make it a subtype of &conditon as you
suggest, but then foreign errors are not a subtype of &error. If a R6RS
wizard is lurking, advice would be helpful.

Anyway, this doesn't need to be in the next draft.