Commits on Nov 17, 2010
  1. Fix the type sanity test in genprimopcode's Syntax.hs

    igfoo committed Nov 17, 2010
    We assume this is what it's supposed to be checking. Certainly the old
    test (t2 == t2) can't be right.
    Spotted by Andres Loeh.
  2. remove -XNoMonomorphismRestriction

    simonmar committed Nov 17, 2010
    This was apparently needed at some point during the new typechecker
    development, but does not seem to be required now.
  3. Document SPECIALISE for imported functions committed Nov 17, 2010
    This is a really useful new facility, but I'd forgotten to document it.
    Pls merge to 7.0 branch
  4. Fix Trac #4498: bang-pattern bindings are monomorphic committed Nov 17, 2010
    This patch forces bang patterns to be monomorphic,
    and documents this fact.
  5. Comments only committed Nov 17, 2010
  6. Fix Trac #4501: a transposition error in DynFlags committed Nov 17, 2010
    Push to STABLE
Commits on Nov 16, 2010
  1. Refactoring of the way that inlinings and rules are activated committed Nov 16, 2010
    Principally, the SimplifierMode now carries several (currently
    four) flags in *all* phases, not just the "Gentle" phase.
    This makes things simpler and more uniform.
    As usual I did more refactoring than I had intended.
    This stuff should go into 7.0.2 in due course, once
    we've checked it solves the DPH performance problems.
  2. Some infrastruture for lambda-lifting committed Nov 16, 2010
    This stuff should have no effect but it sets things
    up so that we can try floating out lambdas of n value
    The new (secret) flag is -ffloatt-lam-args=n.
    This is *not* working yet, but it's got tangled up with
    other stuff I want to commit, and it does no harm.
  3. Occurrence analyser takes account of the phase when handing RULES committed Nov 16, 2010
    See Note [Finding rule RHS free vars]
    This should make Roman happy.
  4. Add -fwarn-lazy-unlifted-bindings to the list of flags committed Nov 16, 2010
  5. Warn a bit less often about unlifted bindings. committed Nov 16, 2010
    Warn when
       (a) a pattern bindings binds unlifted values
       (b) it has no top-level bang
       (c) the RHS has a *lifted* type
    Clause (c) is new, argued for by Simon M
    Eg     x# = 4# + 4#      -- No warning
           (# a,b #) = blah  -- No warning
           I# x = blah       -- Warning
  6. Libraries bytestirng, time, binary need -fno-warn-identities committed Nov 16, 2010
    They all have redundant uses of fromIntegral, but are no under
    GHC HQ control.
  7. Improve documentation for -fwarn-incomplete-patterns committed Nov 16, 2010
  8. Add warning for probable identities (fromIntegral and friends) committed Nov 16, 2010
    See Trac #4488.  The basic idea is to check for
        fun :: ty -> ty
    where fun is one of
      toIntegerName     toRationalName
      fromIntegralName  realToFracName
    There's a (documented) flag to control it -fwarn-identities.
    Currently -Wall switches it on.
  9. Remove unncessary fromIntegral calls committed Nov 16, 2010
Commits on Nov 15, 2010
  1. Fix -fwarn-missing-import-lists (fix Trac #4489) committed Nov 15, 2010
  2. Deriving Typeable changes committed Nov 15, 2010
    * Fix a bug that led to a crash with
        data family T a
        deriving Functor T
    * Allow deriving Typeable for data families
        data family T a
        deriving Typeable1 T
    * Some refactoring and tidying
  3. Ensure that instance overlap errors are report properly committed Nov 15, 2010
    This (annoyingly) requires us to re-flatten the class predicate.
    See Note [Flattening in error message generation]
  4. Ensure that unification variables alloc'd during solving are untouchable committed Nov 15, 2010
    This fixes Trac #4494.  See Note [Extra TcsTv untouchables] in TcSimplify.
Commits on Nov 14, 2010
  1. Remove -Ofile docs

    igfoo committed Nov 14, 2010
    They said:
        (NOTE: not supported since GHC 4.x.  Please ask if you're interested
        in this.)
Commits on Nov 13, 2010
  1. Fix up the ifdefs in Task.c

    igfoo committed Nov 13, 2010
Commits on Nov 12, 2010
  1. Remove use of (now-gone) VarPatOut committed Nov 12, 2010
  2. More modules that need LANGUAGE BangPatterns committed Nov 12, 2010
  3. Add LANGUAGE BangPatterns to modules that use them committed Nov 12, 2010
  4. Warn about top-level bangs (Trac #4477) committed Nov 12, 2010
    A simple refactoring in the parser
Commits on Nov 11, 2010
Commits on Nov 5, 2010
  1. Eventlog: Put correct size for startup event committed Nov 5, 2010
  2. Remove the now-unused constructor VarPatOut committed Nov 5, 2010
  3. Track change in isInlinePragma committed Nov 5, 2010
    In TcBinds we want to use isAnyInlinePragma, to get
    both INLINE and INLINABLE.  I don't know why this isn't
    leading to failures for others!  The (bogus) error I got, 
    triggered by this bug was:
        You cannot SPECIALISE `M.findWithDefault'
          because its definition has no INLINE/INLINABLE pragma