1. 17 Dec, 2004 1 commit
  2. 16 Dec, 2004 1 commit
  3. 15 Dec, 2004 3 commits
  4. 13 Dec, 2004 1 commit
  5. 08 Dec, 2004 3 commits
  6. 06 Dec, 2004 3 commits
    • simonpj's avatar
      [project @ 2004-12-06 10:58:06 by simonpj] · bfb87645
      simonpj authored
      ---------------------
      	Bug in specialisation
      	---------------------
      
      Laszlo managed to get a function like this:
      
      	foo :: Enum a => (# a, Int #)
      
      The specialiser specialised it, resulting in an unboxed tuple
      binding, which Lint objected to.
      
      This commit adds a dummy argument to the specialised function, 
      very like the case for strictness analysis.  For example, at
      type Char we'd get
      
      	foo_char :: State# RealWorld -> (# Char, Int #)
       	foo_char = \_ -> ...
      
      We use a State# type because it generates no argument-passing code 
      at runtime.  (We should really have some other void type for this 
      purpose, because State# is misleading, but this way avoids extra
      types.)
      bfb87645
    • simonpj's avatar
      [project @ 2004-12-06 10:51:36 by simonpj] · f6f3819f
      simonpj authored
      ------------------------------------
      	Bug in loop detection in TcSimplify
      	------------------------------------
      
      The type-class context simplifier has been able to 
      build recursive dictionaries for some time: co-induction.
      That is, you can build a proof for constraint C by assuming
      that C holds when proving the preconditions of C.
      
      You need to be in -fallow-undecidable-instances land to 
      make use of this: see comments with [RECURSIVE DICTIONARIES]
      in TcSimplify.lhs.
      
      Anyway, this is all fine, but I'd implemented it wrong!  You need
      to be very careful with superclasses, or you can make a bogus
      loop by mistake.  This commit fixes it; tests LoopOfTheDay{1,2,3}
      will test it (thanks Ralf Laemmel).
      f6f3819f
    • simonmar's avatar
      [project @ 2004-12-06 10:47:22 by simonmar] · 23843129
      simonmar authored
      Error message fixup
      23843129
  7. 03 Dec, 2004 3 commits
  8. 02 Dec, 2004 4 commits
  9. 01 Dec, 2004 1 commit
  10. 30 Nov, 2004 6 commits
  11. 29 Nov, 2004 7 commits
  12. 26 Nov, 2004 3 commits
    • simonmar's avatar
      [project @ 2004-11-26 16:32:44 by simonmar] · 2dc3b476
      simonmar authored
      Further integration with the new package story.  GHC now supports
      pretty much everything in the package proposal.
      
        - GHC now works in terms of PackageIds (<pkg>-<version>) rather than
          just package names.  You can still specify package names without
          versions on the command line, as long as the name is unambiguous.
      
        - GHC understands hidden/exposed modules in a package, and will refuse
          to import a hidden module.  Also, the hidden/eposed status of packages
          is taken into account.
      
        - I had to remove the old package syntax from ghc-pkg, backwards
          compatibility isn't really practical.
      
        - All the package.conf.in files have been rewritten in the new syntax,
          and contain a complete list of modules in the package.  I've set all
          the versions to 1.0 for now - please check your package(s) and fix the
          version number & other info appropriately.
      
        - New options:
      
      	-hide-package P    sets the expose flag on package P to False
      	-ignore-package P  unregisters P for this compilation
      
      	For comparison, -package P sets the expose flag on package P
              to True, and also causes P to be linked in eagerly.
      
              -package-name is no longer officially supported.  Unofficially, it's
      	a synonym for -ignore-package, which has more or less the same effect
      	as -package-name used to.
      
      	Note that a package may be hidden and yet still be linked into
      	the program, by virtue of being a dependency of some other package.
      	To completely remove a package from the compiler's internal database,
              use -ignore-package.
      
      	The compiler will complain if any two packages in the
              transitive closure of exposed packages contain the same
              module.
      
      	You *must* use -ignore-package P when compiling modules for
              package P, if package P (or an older version of P) is already
              registered.  The compiler will helpfully complain if you don't.
      	The fptools build system does this.
      
         - Note: the Cabal library won't work yet.  It still thinks GHC uses
           the old package config syntax.
      
      Internal changes/cleanups:
      
         - The ModuleName type has gone away.  Modules are now just (a
           newtype of) FastStrings, and don't contain any package information.
           All the package-related knowledge is in DynFlags, which is passed
           down to where it is needed.
      
         - DynFlags manipulation has been cleaned up somewhat: there are no
           global variables holding DynFlags any more, instead the DynFlags
           are passed around properly.
      
         - There are a few less global variables in GHC.  Lots more are
           scheduled for removal.
      
         - -i is now a dynamic flag, as are all the package-related flags (but
           using them in {-# OPTIONS #-} is Officially Not Recommended).
      
         - make -j now appears to work under fptools/libraries/.  Probably
           wouldn't take much to get it working for a whole build.
      2dc3b476
    • simonmar's avatar
      [project @ 2004-11-26 16:30:12 by simonmar] · 67dae2b1
      simonmar authored
      Add missing .hi-boot files
      67dae2b1
    • simonmar's avatar
      [project @ 2004-11-26 16:19:45 by simonmar] · ef5b4b14
      simonmar authored
      Further integration with the new package story.  GHC now supports
      pretty much everything in the package proposal.
      
        - GHC now works in terms of PackageIds (<pkg>-<version>) rather than
          just package names.  You can still specify package names without
          versions on the command line, as long as the name is unambiguous.
      
        - GHC understands hidden/exposed modules in a package, and will refuse
          to import a hidden module.  Also, the hidden/eposed status of packages
          is taken into account.
      
        - I had to remove the old package syntax from ghc-pkg, backwards
          compatibility isn't really practical.
      
        - All the package.conf.in files have been rewritten in the new syntax,
          and contain a complete list of modules in the package.  I've set all
          the versions to 1.0 for now - please check your package(s) and fix the
          version number & other info appropriately.
      
        - New options:
      
      	-hide-package P    sets the expose flag on package P to False
      	-ignore-package P  unregisters P for this compilation
      
      	For comparison, -package P sets the expose flag on package P
              to True, and also causes P to be linked in eagerly.
      
              -package-name is no longer officially supported.  Unofficially, it's
      	a synonym for -ignore-package, which has more or less the same effect
      	as -package-name used to.
      
      	Note that a package may be hidden and yet still be linked into
      	the program, by virtue of being a dependency of some other package.
      	To completely remove a package from the compiler's internal database,
              use -ignore-package.
      
      	The compiler will complain if any two packages in the
              transitive closure of exposed packages contain the same
              module.
      
      	You *must* use -ignore-package P when compiling modules for
              package P, if package P (or an older version of P) is already
              registered.  The compiler will helpfully complain if you don't.
      	The fptools build system does this.
      
         - Note: the Cabal library won't work yet.  It still thinks GHC uses
           the old package config syntax.
      
      Internal changes/cleanups:
      
         - The ModuleName type has gone away.  Modules are now just (a
           newtype of) FastStrings, and don't contain any package information.
           All the package-related knowledge is in DynFlags, which is passed
           down to where it is needed.
      
         - DynFlags manipulation has been cleaned up somewhat: there are no
           global variables holding DynFlags any more, instead the DynFlags
           are passed around properly.
      
         - There are a few less global variables in GHC.  Lots more are
           scheduled for removal.
      
         - -i is now a dynamic flag, as are all the package-related flags (but
           using them in {-# OPTIONS #-} is Officially Not Recommended).
      
         - make -j now appears to work under fptools/libraries/.  Probably
           wouldn't take much to get it working for a whole build.
      ef5b4b14
  13. 25 Nov, 2004 1 commit
    • simonpj's avatar
      [project @ 2004-11-25 11:36:34 by simonpj] · 1f7da302
      simonpj authored
      ------------------------------------------
      	Keep-alive set and Template Haskell quotes
      	------------------------------------------
      
      a) Template Haskell quotes should be able to mention top-leve
         things without resorting to lifting.  Example
      
      	module Foo( foo ) where
      	  f x = x
      	  foo = [| f 4 |]
      
         Here the reference to 'f' is ok; no need to 'lift' it.
         The relevant changes are in TcExpr.tcId
      
      b) However, we must take care not to discard the binding for f,
         so we add it to the 'keep-alive' set for the module.  I've
         now made this into (another) mutable bucket, tcg_keep, 
         in the TcGblEnv
      
      c) That in turn led me to look at the handling of orphan rules;
         as a result I made IdCoreRule into its own data type, which
         has simle but non-local ramifications
      1f7da302
  14. 24 Nov, 2004 1 commit
    • simonmar's avatar
      [project @ 2004-11-24 13:03:26 by simonmar] · c1a7d1a6
      simonmar authored
      Update s/r conflict total after Ross added four more.  The total
      according to the comment is currently one more than the real number,
      so one of them must have gone away.  At some point we should track
      down which one.
      c1a7d1a6
  15. 23 Nov, 2004 1 commit
  16. 21 Nov, 2004 1 commit
    • desrt's avatar
      [project @ 2004-11-21 22:25:23 by desrt] · a1947dc7
      desrt authored
      rtsconfig, smrep: fixes to not put info next to code (like ia64)
      machregs: define registers for powerpc64 - same as ppc32
      tailcalls: for now do tailcalls as normal calls. (ok on powerpc since
                 calls don't grow the stack and we don't care about LR anyway)
      a1947dc7