Re: [GHC] #12063: Knot-tying failure when type-synonym refers to non-existent data