Re: [GHC] #4309: Painfully large errors with silly GADT instances