Current source relies on odd (perhaps broken) behaviour of (an older release of) ca65 #10
Description
See cc65/cc65#2208
The current source does some (rather complex and hard to follow) magic things to figure out the location of certain variables. This broke due to a recent fix in ca65 (arguable the behaviour was broken before AND after the fix - but its now broken in a different way, while another case was fixed).
However, to properly deal with this, the source should be changed to use less magic instead, perhaps use different linker configs for each target binary.
Obviously the problem should ideally be fixed in ca65 as well - but i don't see that happening anytime soon (it's really not trivial to do). I don't see that recent fix being reverted either for that matter, as it fixed a much more relevant problem.