panic! mkWWcpr: not a product

[nobody]

I encountered the following bug:

ghc-6.4: panic! (the `impossible' happened, GHC version
6.4):
        mkWWcpr: not a product SPIRziSPIR.SPIR{tc r3jI}

Please report it as a compiler bug to
glasgow-haskell-bugs@haskell.org,
or http://sourceforge.net/projects/ghc/.

This was in conjunction with the use of .hs-boot files
for recursive modules, though I don't know for sure if that
actually caused the problem.

Unfortunate I can't provide a simple reproducible test case
at this point, but I thought I'd let you know about the
bug anyway.

Fergus Henderson <fjh-mailbox-30@galois.com>

[simonpj]

Logged In: YES 
user_id=50165

Definitely a bug, but hard to find without a test case.

Were you using --make?  We've got a known problem there.

[simonpj]

Logged In: YES 
user_id=50165

Definitely a bug, but hard to find without a test case.

Were you using --make?  We've got a known problem there.

[fergus]

Logged In: YES 
user_id=135331

Yes, I was using --make.

[guest]

I have the same problem with ghc 6.4.1.

I was not using --make, but using -O2 -prof -auto-all.

Here is the source:

module M where

newtype T a = T { unT :: a }

f = unT

test cs = f $ case cs of
  [] -> T []
  (x:xs) -> T $ test cs

$ ghc -c M.hs -O2 -prof -auto-all
ghc-6.4.1: panic! (the `impossible' happened, GHC version 6.4.1):
    mkWWcpr: not a product [a{tv a19n}]

Please report it as a compiler bug to glasgow-haskell-bugs@haskell.org,
or http://sourceforge.net/projects/ghc/.

[simonpj]

The nice small test case given by 'guest' was great. It exposed a long-standing bug (2002 at least). Now fixed; the fix will be in 6.4.2.

There remains a harder-to-fix bug involving recursive modules and hi-boot files, so I'm not actually going to close it, alas.

Simon

[simonmar]

dup of #930