Possible bug in Renamer when dealing with orphans

In a branch of GHC with Growable ASTs, the code compiles with devel2 configuration, but fails with ./validate. It passes all the test cases, except two that already fail in HEAD (see #14126).

Especifically, the compiler returns the following error:

ghc-stage1: panic! (the 'impossible' happened)
  (GHC version 8.3.20170806 for x86_64-unknown-linux):
	ASSERT failed!
  HsPat [GHC.Base, GHC.Float, Data.Binary.Generic,
         Data.ByteString.Builder, HsBinds, HsDoc, HsExpr, HsLit, HsPat,
         HsTypes, PprCore, GHC.LanguageExtensions,
         Data.Time.Calendar.Gregorian, Data.Time.Format.Parse,
         Data.Time.LocalTime.Internal.LocalTime,
         Data.Time.LocalTime.Internal.ZonedTime]
  Call stack:
      CallStack (from HasCallStack):
        prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1133:58 in ghc:Outputable
        callStackDoc, called at compiler/utils/Outputable.hs:1188:22 in ghc:Outputable
        assertPprPanic, called at compiler/rename/RnNames.hs:364:99 in ghc:RnNames
  Call stack:
      CallStack (from HasCallStack):
        prettyCurrentCallStack, called at compiler/utils/Outputable.hs:1133:58 in ghc:Outputable
        callStackDoc, called at compiler/utils/Outputable.hs:1137:37 in ghc:Outputable
        pprPanic, called at compiler/utils/Outputable.hs:1186:5 in ghc:Outputable
        assertPprPanic, called at compiler/rename/RnNames.hs:364:99 in ghc:RnNames

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

I suspect (as hinted above) it has to do with a bug in the renamer when dealing with orphan instances (and probably in conjunction with boot files).

Trac metadata

Trac field	Value
Version	8.2.1
Type	Bug
TypeOfFailure	OtherFailure
Priority	normal
Resolution	Unresolved
Component	Compiler (Type checker)
Test case
Differential revisions
BlockedBy
Related
Blocking
CC	alanz, simonpj
Operating system
Architecture