GitLab

cf Trac #5509

It now knows about all the differences between the extensions that
GHC knows, and the extensions that Cabal knows.

It was failing due to quoting differences between Windows and other
platforms.

This reverts commit e82472fd68db55a23f07d395fa8104eb6a1e1cc0.

This leads to gratuitous Windows/Unix differences
which aren't germane to the test

Test cases for GHCi data, type, newtype, class, instance,
deriving, etc declarations.

The OS X codegen has higher memory requirements than the elf codegen.

Integers now take up less space, so space usage has improved in this
Integer-heavy test.

In my dev tree, this currently fails with:

    ghc-stage2: panic! (the 'impossible' happened)
      (GHC version 7.3.20110916 for x86_64-unknown-linux):
            Evaluated the place holder for mkInteger

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

It turns out that the new treatment of Integer literals
means that the debugger can see an Integer value more often
than before (it's a value not a thunk). That improves the
debugger output; hence all these diffs.