1. 30 Sep, 2016 2 commits
    • Simon Peyton Jones's avatar
      Fix impredicativity (again) · b612da66
      Simon Peyton Jones authored
      This patch fixes Trac #12616.
      
      Dignosis.  In TcUnify.tc_sub_type_ds we were going to some trouble to
      support co- and contra-variance even for impredicative types.  With
      -XImpredicativeTYpes, this allowed a unification variable to be
      unified with a polytype (probably wrongly) and that caused later
      trouble in the constraint solver, where -XImpredicativeTypes was /not/
      on.  In effect, -XImpredicativeTypes can't be switched on locally.
      
      Why did we want ImpredicativeTypes locally?  Because the program
      generated by GND for a higher-rank method involved impredicative
      instantation of 'coerce':
            op = coerce op   -- where op has a higher rank type
      See Note [Newtype-deriving instances] in TcGenDeriv.
      
      Cure.
      
      1.  It is ghastly to rely on ImpredicativeTypes (a 100% flaky
          feature) to instantiate coerce polymorphically.  Happily we
          now have Visible Type Application, so I've used that instead
          which should be solid and reliable.
      
      2.  I deleted the code in tc_sub_type_ds that allows the constraint
          solver to "look through" a unification variable to find a
          polytype.  That used to be essential in the days of ReturnTv,
          but it's utterly unreliable and should be consigned to the dustbin
          of history.  (We have ExpType now for the essential uses.)
      
      Tests involving ImpredicativeTypes are affected, but I'm not worried
      about them... it's advertised as a feature you can't rely on, and
      I want to reform it outright.
      b612da66
    • Simon Peyton Jones's avatar
      Fix a bug in occurs checking · 66a8c194
      Simon Peyton Jones authored
      1. Trac #12593 exposed a long-standing bug in the occurs
         checking machinery.  When unifying two type variables
                a ~ b
         where a /= b, we were assuming that there could be
         no occurs-check error.  But there can: 'a' can occur
         in b's kind!  When the RHS was a non-tyvar we used
         occurCheckExpand, which /did/ look in kinds, but not
         when the RHS was a tyvar.
      
         This bug has been lurking ever since TypeInType, maybe
         longer.  And it was present both in TcUnify (the on-the-fly
         unifier), and in TcInteract.
      
         I ended up refactoring both so that the tyvar/tyvar
         path naturally goes through the same occurs-check as
         non-tyvar rhss.  It's simpler and more robust now.
      
         One good thing is that both unifiers now share
             TcType.swapOverVars
             TcType.canSolveByUnification
         previously they had different logic for the same goals
      
      2. Fixing this bug exposed another!  In T11635 we end
         up unifying
         (alpha :: forall k. k->*) ~ (beta :: forall k. k->*)
         Now that the occurs check is done for tyvars too, we
         look inside beta's kind.  And then reject the program
         becuase of the forall inside there.  But in fact that
         forall is fine -- it does not count as impredicative
         polymoprhism.   See Note [Checking for foralls]
         in TcType.
      
      3. All this fuss around occurrence checking forced me
         to look at TcUnify.checkTauTvUpdate
                and TcType.occurCheckExpand
         There's a lot of duplication there, and I managed
         to elminate quite a bit of it. For example,
         checkTauTvUpdate called a local 'defer_me'; and then
         called occurCheckExpand, which then used a very
         similar 'fast_check'.
      
         Things are better, but there is more to do.
      66a8c194
  2. 26 Sep, 2016 1 commit
  3. 12 Sep, 2016 1 commit
    • Simon Peyton Jones's avatar
      Be less picky about reporing inaccessible code · 03541cba
      Simon Peyton Jones authored
      Triggered by the discussion on Trac #12466, this patch
      makes GHC less aggressive about reporting an error when
      there are insoluble Givens.
      
      Being so agressive was making some libraries fail to
      compile, and is arguably wrong in at least some cases.
      See the discussion on the ticket.
      
      Several test now pass when they failed before; see
      the files-modified list for this patch.
      03541cba
  4. 31 Aug, 2016 2 commits
  5. 21 Aug, 2016 1 commit
    • Edward Z. Yang's avatar
      Axe initIfaceTc, tie the knot through HPT (or if_rec_types). · e907e1f1
      Edward Z. Yang authored
      
      
      Summary:
      initIfaceTc was originally used to make sure when we typecheck
      an interface, it can find the TyThings for things it itself
      defined.  However, in the case of retypecheckLoop, this wasn't
      necessary because we ALREADY tied the knot through the HPT.
      
      This commit removes initIfaceTc, instead relying on the HPT
      to tie the knot.  genModDetails' caller needed to be modified
      to tie the knot, but there are not that many call-sites of
      typecheckIface so the change is quite reasonable.
      
      We also introduce a new 'initIfaceLoad', which does
      NOT set up 'if_rec_types'.  It's used when we're
      typechecking old, up-to-date interfaces in, since we're
      never going to update the type environment.
      
      The full details are in Note [Knot-tying typecheckIface].
      Displeasingly, we need a special case to handle DFuns in
      the case of tcHiBootIface, see
      Note [DFun knot-tying special case] for the gory details.
      
      I also added another test which tickles a bug in a buggy
      version of this patch (see "Why the seq?")
      Signed-off-by: default avatarEdward Z. Yang <ezyang@cs.stanford.edu>
      
      Test Plan: validate
      
      Reviewers: simonpj, austin, bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2349
      e907e1f1
  6. 21 Jul, 2016 1 commit
  7. 20 Jul, 2016 2 commits
  8. 14 Jul, 2016 1 commit
  9. 05 Jul, 2016 1 commit
  10. 29 Jun, 2016 1 commit
  11. 28 Jun, 2016 1 commit
    • Simon Peyton Jones's avatar
      Deal correctly with unused imports for 'coerce' · 23b80ac4
      Simon Peyton Jones authored
      We only do newtype unwrapping for Coercible constraints if
      the newtype's data constructor is in scope.  We were trying to
      record the fact that the data constructor was thereby 'used', so
      that an import statement would not be flagged as unnecsssary
      (by -Wunused-imports).
      
      But the code was simply wrong. It was wrong because it assumed
      that only one level of unwrapping happened, whereas
      tcTopNormaliseNewTypeTF_maybe actually unwraps multiple layers.
      So we need to return a /list/ of data constructors that are used.
      
      This entailed a bit of refactoring, as usual.
      
      Fixes Trac #12067
      23b80ac4
  12. 23 Jun, 2016 2 commits
  13. 21 Jun, 2016 1 commit
  14. 20 Jun, 2016 4 commits
  15. 15 Jun, 2016 1 commit
    • Simon Peyton Jones's avatar
      Major patch to introduce TyConBinder · e368f326
      Simon Peyton Jones authored
      Before this patch, following the TypeInType innovations,
      each TyCon had two lists:
        - tyConBinders :: [TyBinder]
        - tyConTyVars  :: [TyVar]
      
      They were in 1-1 correspondence and contained
      overlapping information.  More broadly, there were many
      places where we had to pass around this pair of lists,
      instead of a single list.
      
      This commit tidies all that up, by having just one list of
      binders in a TyCon:
      
        - tyConBinders :: [TyConBinder]
      
      The new data types look like this:
      
        Var.hs:
           data TyVarBndr tyvar vis = TvBndr tyvar vis
           data VisibilityFlag = Visible | Specified | Invisible
           type TyVarBinder = TyVarBndr TyVar VisibilityFlag
      
        TyCon.hs:
           type TyConBinder = TyVarBndr TyVar TyConBndrVis
      
           data TyConBndrVis
             = NamedTCB VisibilityFlag
             | AnonTCB
      
        TyCoRep.hs:
           data TyBinder
             = Named TyVarBinder
             | Anon Type
      
      Note that Var.TyVarBdr has moved from TyCoRep and has been
      made polymorphic in the tyvar and visiblity fields:
      
           type TyVarBinder = TyVarBndr TyVar VisibilityFlag
              -- Used in ForAllTy
           type TyConBinder = TyVarBndr TyVar TyConBndrVis
              -- Used in TyCon
      
           type IfaceForAllBndr  = TyVarBndr IfaceTvBndr VisibilityFlag
           type IfaceTyConBinder = TyVarBndr IfaceTvBndr TyConBndrVis
               -- Ditto, in interface files
      
      There are a zillion knock-on changes, but everything
      arises from these types.  It was a bit fiddly to get the
      module loops to work out right!
      
      Some smaller points
      ~~~~~~~~~~~~~~~~~~~
      * Nice new functions
          TysPrim.mkTemplateKiTyVars
          TysPrim.mkTemplateTyConBinders
        which help you make the tyvar binders for dependently-typed
        TyCons.  See comments with their definition.
      
      * The change showed up a bug in TcGenGenerics.tc_mkRepTy, where the code
        was making an assumption about the order of the kind variables in the
        kind of GHC.Generics.(:.:).  I fixed this; see TcGenGenerics.mkComp.
      e368f326
  16. 14 Jun, 2016 1 commit
  17. 13 Jun, 2016 1 commit
    • Simon Peyton Jones's avatar
      Improve typechecking of let-bindings · 15b9bf4b
      Simon Peyton Jones authored
      This major commit was initially triggered by #11339, but it spiraled
      into a major review of the way in which type signatures for bindings
      are handled, especially partial type signatures.  On the way I fixed a
      number of other bugs, namely
         #12069
         #12033
         #11700
         #11339
         #11670
      
      The main change is that I completely reorganised the way in which type
      signatures in bindings are handled. The new story is in TcSigs
      Note [Overview of type signatures].  Some specific:
      
      * Changes in the data types for signatures in TcRnTypes:
        TcIdSigInfo and new TcIdSigInst
      
      * New module TcSigs deals with typechecking type signatures
        and pragmas. It contains code mostly moved from TcBinds,
        which is already too big
      
      * HsTypes: I swapped the nesting of HsWildCardBndrs
        and HsImplicitBndsrs, so that the wildcards are on the
        oustide not the insidde in a LHsSigWcType.  This is just
        a matter of convenient, nothing deep.
      
      There are a host of other changes as knock-on effects, and
      it all took FAR longer than I anticipated :-).  But it is
      a significant improvement, I think.
      
      Lots of error messages changed slightly, some just variants but
      some modest improvements.
      
      New tests
      
      * typecheck/should_compile
          * SigTyVars: a scoped-tyvar test
          * ExPat, ExPatFail: existential pattern bindings
          * T12069
          * T11700
          * T11339
      
      * partial-sigs/should_compile
          * T12033
          * T11339a
          * T11670
      
      One thing to check:
      
      * Small change to output from ghc-api/landmines.
        Need to check with Alan Zimmerman
      15b9bf4b
  18. 09 Jun, 2016 1 commit
  19. 04 May, 2016 1 commit
    • niteria's avatar
      Kill non-deterministic foldUFM in TrieMap and TcAppMap · ad4392c1
      niteria authored
      Summary:
      foldUFM introduces unnecessary non-determinism that actually
      leads to different generated code as explained in
      Note [TrieMap determinism].
      
      As we're switching from UniqFM to UniqDFM here you might be
      concerned about performance. There's nothing that ./validate
      detects. nofib reports no change in Compile Allocations, but
      Compile Time got better on some tests and worse on some,
      yielding this summary:
      
              -1 s.d.                -----            -3.8%
              +1 s.d.                -----            +5.4%
              Average                -----            +0.7%
      
      This is not a fair comparison as the order of Uniques
      changes what GHC is actually doing. One benefit from making
      this deterministic is also that it will make the
      performance results more stable.
      
      Full nofib results: P108
      
      Test Plan: ./validate, nofib
      
      Reviewers: goldfire, simonpj, simonmar, austin, bgamari
      
      Reviewed By: simonpj
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2169
      
      GHC Trac Issues: #4012
      ad4392c1
  20. 26 Apr, 2016 2 commits
  21. 22 Apr, 2016 2 commits
    • Simon Peyton Jones's avatar
      Warn about simplifiable class constraints · 9421b0c7
      Simon Peyton Jones authored
      Provoked by Trac #11948, this patch adds a new warning to GHC
      
        -Wsimplifiable-class-constraints
      
      It warns if you write a class constraint in a type signature that
      can be simplified by an existing instance declaration.  Almost always
      this means you should simplify it right now; type inference is very
      fragile without it, as #11948 shows.
      
      I've put the warning as on-by-default, but I suppose that if there are
      howls of protest we can move it out (as happened for -Wredundant-constraints.
      
      It actually found an example of an over-complicated context in CmmNode.
      
      Quite a few tests use these weird contexts to trigger something else,
      so I had to suppress the warning in those.
      
      The 'haskeline' library has a few occurrences of the warning (which
      I think should be fixed), so I switched it off for that library in
      warnings.mk.
      
      The warning itself is done in TcValidity.check_class_pred.
      
      HOWEVER, when type inference fails we get a type error; and the error
      suppresses the (informative) warning.  So as things stand, the warning
      only happens when it doesn't cause a problem.  Not sure what to do
      about this, but this patch takes us forward, I think.
      9421b0c7
    • Simon Peyton Jones's avatar
      Do not use defaulting in ambiguity check · edf54d72
      Simon Peyton Jones authored
      This fixes Trac #11947.  See TcSimplify
      Note [No defaulting in the ambiguity check]
      edf54d72
  22. 20 Apr, 2016 1 commit
    • Simon Peyton Jones's avatar
      SCC analysis for instances as well as types/classes · 353d8ae6
      Simon Peyton Jones authored
      This big patch is in pursuit of Trac #11348.
      
      It is largely the work of Alex Veith (thank you!), with some
      follow-up simplification and refactoring from Simon PJ.
      
      The main payload is described in RnSource
        Note [Dependency analysis of type, class, and instance decls]
      which is pretty detailed.
      
      * There is a new data type HsDecls.TyClGroup, for a strongly
        connected component of type/class/instance/role decls.
      
        The hs_instds field of HsGroup disappears, in consequence
      
        This forces some knock-on changes, including a minor
        haddock submodule update
      
      Smaller, weakly-related things
      
      * I found that both the renamer and typechecker were building an
        identical env for RoleAnnots, so I put common code for
        RoleAnnotEnv in RnEnv.
      
      * I found that tcInstDecls1 had very clumsy error handling, so I
        put it together into TcInstDcls.doClsInstErrorChecks
      353d8ae6
  23. 15 Apr, 2016 1 commit
  24. 12 Apr, 2016 1 commit
    • eir@cis.upenn.edu's avatar
      Fix #11811. · b1084fd7
      eir@cis.upenn.edu authored
      Previously, I had forgotten to omit variables already in scope
      from the TypeInType CUSK check. Simple enough to fix.
      
      Test case: typecheck/should_compile/T11811
      b1084fd7
  25. 05 Apr, 2016 1 commit
    • Herbert Valerio Riedel's avatar
      Fix misattribution of `-Wunused-local-binds` warnings · 1e6ec124
      Herbert Valerio Riedel authored
      This fixes a bug where warnings actually controlled by
      
      - `Opt_WarnUnusedMatches`
      - `Opt_WarnUnusedTypePatterns`
      - `Opt_WarnUnusedTopBinds`
      
      were incorrectly reported as being controlled by
      `Opt_WarnUnusedLocalBinds` as well
      
      This bug was introduced in bb5afd3c
      while implementing #10752
      
      Test Plan: ./validate still running -- testsuite output wiggles expected
      
      Reviewers: barrucadu, quchen, austin, bgamari
      
      Reviewed By: bgamari
      
      Subscribers: thomie
      
      Differential Revision: https://phabricator.haskell.org/D2077
      1e6ec124
  26. 30 Mar, 2016 1 commit
  27. 26 Mar, 2016 1 commit
  28. 17 Mar, 2016 1 commit
  29. 15 Mar, 2016 2 commits
  30. 29 Feb, 2016 1 commit