GitLab

(cherry picked from commit 59b4e6d0)

The constraints for Functor don't line up 1-1 with the arguments
(they are fetched out from sub-terms of the type), but the surrounding
code was mistakenly assuming they were in 1-1 association.

(cherry picked from commit 13a330e8)

Conflicts:
	compiler/typecheck/TcDeriv.lhs

This bug was causing Trac #9199

(cherry picked from commit 571f0adc)

Conflicts:
	testsuite/tests/th/all.T

Signed-off-by: Austin Seipp <austin@well-typed.com>

This reverts commit f6e49423

.
Signed-off-by: Austin Seipp <austin@well-typed.com>

(cherry picked from commit 1b45c6b4959762209dbcabf3290feb8673a60b23)

The (pure) unifier tcUnifyTys returns an idempotent substitution.
But previously the kinds of type variables free in the range of
the subst could have un-substituted kind variables.

This patch fixes that, fixing Trac #9106.

See Note [Finding the substitution fixpoint] in Unify

(cherry picked from commit d8d97113)

This fixes Trac #9096

(cherry picked from commit 770e16fc)

In a special case for trivial RHSs (see DmdAnal.unpackTrivial),
I'd forgotten to include a demand for the RHS itself.
See Note [Remember to demand the function itself].

Thanks to David Terei for guiding me to the bug,
at PLDI in Edinburgh.

(cherry picked from commit 7d9feb26)

See Note [Un-warnable import decls] in RnNames.

Fixes Trac #9061.

(cherry picked from commit 9f3e39d5)

Signed-off-by: Austin Seipp <austin@well-typed.com>
(cherry picked from commit c226d25f)

Conflicts:
	testsuite/tests/th/all.T

(cherry picked from commit 2a463ebe)

A previous fix to this was wrong: f5879acd
and left some unreachable code behind.  So rather than try to be clever and
do this at the same time as the strongly-connected-component analysis, I'm
doing a separate reachability pass first.

(cherry picked from commit e577a523)

(cherry picked from commit 96a95f05)

This fixes #9078

.
Signed-off-by: Austin Seipp <austin@well-typed.com>
(cherry picked from commit b75d126e)

(cherry picked from commit fc0ed8a7)

Conflicts:
	testsuite/tests/codeGen/should_run/all.T

The provided tests test both annotation generation and reification
from Template Haskell.  Both --make and compilation via separate
units (ghc -c) are tested.
Signed-off-by: Austin Seipp <austin@well-typed.com>
(cherry picked from commit 5f5e326c)

(cherry picked from commit cc3ccf9f)

Conflicts:
	testsuite/tests/deriving/should_compile/all.T

The new stg_gc_prim_p_ll stack frame was missing an info table.  This
is a regression since 7.6, because this stuff was part of a cleanup
that happened in 7.7.

(cherry picked from commit ab8bb489)

(cherry picked from commit ff9f9a7f)

This fixes Trac #8954.

There were actually three places where tuple occ-names
were parsed:
  - IfaceEnv.lookupOrigNameCache
  - Convert.isBuiltInOcc
  - OccName.isTupleOcc_maybe

I combined all three into TysWiredIn.isBuiltInOcc_maybe
Much nicer.

(cherry picked from commit 750271e6)

Conflicts:
	testsuite/tests/th/all.T

This was just an omission, which showed up as Trac #8966

(cherry picked from commit 4dc9f986)

(cherry picked from commit d8d798b1)

Conflicts:
	testsuite/tests/th/all.T

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>
(cherry picked from commit c7498bbd)

In tidying up the flattener I introduced an error that no
regression test caught, giving rise to Trac #8978, #8979.
It shows up if you have a type synonym whose RHS mentions
type functions, such sas
     type family F a
     type T a = (F a, a)   -- This synonym isn't properly flattened

The fix is easy, but sadly the bug is in the released GHC 7.8.1

(cherry picked from commit b8132a9d)

Signed-off-by: Austin Seipp <austin@well-typed.com>

Gergő Érdi authored Apr 06, 2014 and thoughtpolice committed Apr 07, 2014

(#8961)

(cherry picked from commit 8f831ec5)

Simon Peyton Jones authored Apr 03, 2014 and thoughtpolice committed Apr 07, 2014

A palpable bug, although one that will rarely bite

(cherry picked from commit 3671d002)

Simon Peyton Jones authored Apr 03, 2014 and thoughtpolice committed Apr 07, 2014

(cherry picked from commit 8bf8ce1e)

eir@cis.upenn.edu authored Apr 05, 2014 and thoughtpolice committed Apr 07, 2014

We now do role inference on stupid datatype contexts, allowing a
lightweight role annotation syntax.

(cherry picked from commit d468cd37)

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

There was a special case in rewriteEvidence, looking like:
  = return (Just (if ctEvPred old_ev `tcEqType` new_pred
                  then old_ev
                  else old_ev { ctev_pred = new_pred }))

But this was wrong: old_pred and new_pred might differ in the kind
of a TyVar occurrence, in which case tcEqType would not notice,
but we really, really want new_pred.  This caused Trac #8913.

I solved this by dropping the whole test, and instead making
the flattener preserve type synonyms. This was easy because
TcEvidence has TcTyConAppCo which (unlike) Coercion, handles
synonyms.

(cherry picked from commit 6ae678e3)

Trac #8912 pointed out that GHC 7.4 and 7.6 have omitted this test, although
7.2 and earlier had it.  This patch puts the test back in, and refactors a
little.

(cherry picked from commit a8b7b28c)

Conflicts:
	testsuite/tests/typecheck/should_fail/all.T

The fix was to ghc-prim.

(cherry picked from commit 16d04d90)

See #8827 - for now, we're making GND unsafe again.

This also fixes the tests since they were originally not using the new
unicode quote style we're using.

This reverts commit a8a01e74.

(cherry picked from commit 8f730377)

Simon Peyton Jones authored Mar 23, 2014 and thoughtpolice committed Mar 23, 2014

(cherry picked from commit 7973bfb8)

Simon Peyton Jones authored Mar 17, 2014 and thoughtpolice committed Mar 23, 2014

(cherry picked from commit 0e2155dd)

Simon Peyton Jones authored Mar 07, 2014 and thoughtpolice committed Mar 23, 2014

(cherry picked from commit 062391be)

Simon Peyton Jones authored Mar 07, 2014 and thoughtpolice committed Mar 23, 2014

The issue here is described in Note [Binding scoped type variables] in
TcPat.  When implementing this fix I was able to make things quite a
bit simpler:
 * The type variables in a type signature now never unify
   with each other, and so can be straightfoward skolems.
 * We only need the SigTv stuff for signatures in patterns,
   and for kind variables.

(cherry picked from commit cf1a0f97)