Does this patch make it so that remaps are considered for header-line even if header-line-active no longer inherits from header-line? If so, I'd consider that even more surprising. The more I think about this, the less inclined I'd be to play special case whack-a-mole with it and the more I'd be inclined to either "live with it" or figure out a way to disable remapping entirely in certain rendering contexts. Tab bar mode is another one that comes to mind that probably shouldn't use remaps at all when rendering. We could treat the computing of basic faces similarly… aside from the initial remap of *-inactive/active, remapping would stop being considered. If that's not feasible, then I don't see this proposed patch as better. Users have already made do with the header-line transition in 29 it sounds like.

Thanks,