- Cvs-nhc98 - Haskell.org

patch applied (hat): Moved trace import magic into SysDeps
by Malcolm Wallace 10 Oct '06

10 Oct '06

Sun Jun 20 05:00:53 PDT 2004 panne * Moved trace import magic into SysDeps M ./src/compiler98/Extra.hs -6 +1 M ./src/compiler98/SysDeps.hs -1 +9

1 0

patch applied (hat): Start concentrating system dependencies into a single place
by Malcolm Wallace 10 Oct '06

10 Oct '06

Sun Jun 20 04:56:55 PDT 2004 panne * Start concentrating system dependencies into a single place (PackedString vs. Data.PackedString only for now) M ./src/compiler98/Case.hs -1 +1 M ./src/compiler98/CaseLib.hs -1 +1 M ./src/compiler98/Export.hs -1 +1 M ./src/compiler98/FFITrans.hs -1 +1 M ./src/compiler98/Fixity.hs -1 +1 M ./src/compiler98/Foreign.hs -1 +1 M ./src/compiler98/GcodeFix.hs -1 +1 M ./src/compiler98/IExtract.hs -1 +1 M ./src/compiler98/Import.hs -1 +1 M ./src/compiler98/ImportState.hs -1 +1 M ./src/compiler98/Info.hs -1 +1 M ./src/compiler98/IntState.hs -1 +1 M ./src/compiler98/LexLow.hs -1 +1 M ./src/compiler98/LexPre.hs -1 +1 M ./src/compiler98/Lexical.hs -1 +1 M ./src/compiler98/MainNew.hs -1 +1 M ./src/compiler98/Makefile -1 +1 M ./src/compiler98/Need.hs -1 +1 M ./src/compiler98/Nice.hs -1 +1 M ./src/compiler98/Parse2.hs -1 +1 M ./src/compiler98/ParseLib.hs -1 +1 M ./src/compiler98/PreImport.hs -1 +1 M ./src/compiler98/PrettySyntax.hs -1 +1 M ./src/compiler98/PrimCode.hs -1 +1 M ./src/compiler98/Rename.hs -1 +1 M ./src/compiler98/RenameLib.hs -1 +1 M ./src/compiler98/ReportImports.hs -1 +1 M ./src/compiler98/RmClasses.hs -1 +1 M ./src/compiler98/STGArity.hs -1 +1 M ./src/compiler98/Syntax.hs -1 +1 A ./src/compiler98/SysDeps.hs M ./src/compiler98/TokenId.hs -1 +1 M ./src/compiler98/Type.hs -1 +1 M ./src/compiler98/TypeUnify.hs -1 +1 M ./src/compiler98/TypeUtil.hs -1 +1

1 0

patch applied (hat): When a newtype is exported abstractly, e.g. CInt(), write the
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri Jun 18 15:46:20 PDT 2004 malcolm * When a newtype is exported abstractly, e.g. CInt(), write the constructor and renamed type into the interface file regardless. This is because FFI'd code (and eventually, deriving clauses) need to know the original type. (Fixes uses IEsome and IEsel to ensure that constructor is not actually visible to an importing module, but can still be re-exported.) M ./src/compiler98/IExtract.hs -1 +6 M ./src/compiler98/Info.hs -1 +5 M ./src/compiler98/RenameLib.hs -4 +5

1 0

patch applied (hat): Simplify re-ordering of module imports. Need to ensure that NHC.Internal
by Malcolm Wallace 10 Oct '06

10 Oct '06

Fri Jun 18 03:41:59 PDT 2004 malcolm * Simplify re-ordering of module imports. Need to ensure that NHC.Internal comes last with Prelude, but there is no need to sort the other modules lexicographically. M ./src/compiler98/PreImport.hs -3 +13

1 0

patch applied (hat): Improve error message only.
by Malcolm Wallace 10 Oct '06

10 Oct '06

Wed Jun 16 07:30:05 PDT 2004 malcolm * Improve error message only. M ./src/compiler98/Type.hs -2 +2

1 0

patch applied (hat): Fix a pattern-matching bug. Consider the following example:
by Malcolm Wallace 10 Oct '06

10 Oct '06

Wed Jun 16 07:23:09 PDT 2004 malcolm * Fix a pattern-matching bug. Consider the following example: ------------------------------------------------ module Foo where newtype Blah = Blah Int deriving (Eq, Show) instance Num Blah where (Blah i) + (Blah j) = Blah (i + j) ... fromInteger x = Blah (fromInteger x) bar :: Blah -> Bool bar 0 = False bar _ = True ------------------------------------------------ nhc98 previously failed with: Fail: What? matchAltIf at 15:9 because the dictionary for (Num.fromInteger {dict:Blah} 0) in the pattern-match was being flattened to (Num.Blah.fromInteger 0), which made it impossible to later choose the right instance of (Eq.==). This fix ensures that fsExp knows whether it is in an expression context or a pattern-context, and for patterns it now refuses to flatten the dictionary. M ./src/compiler98/FixSyntax.hs -71 +69

1 0

patch applied (hat): Warning police
by Malcolm Wallace 10 Oct '06

10 Oct '06

Tue Jun 15 09:39:44 PDT 2004 panne * Warning police M ./src/compiler98/Foreign.hs -65 +85

1 0

patch applied (hat): Eliminate warnings of
by Malcolm Wallace 10 Oct '06

10 Oct '06

Mon Jun 14 08:12:20 PDT 2004 malcolm * Eliminate warnings of foreign import/export has non-primitive type: Prelude.-> for foreign import "wrapper". It is correct that the first argument of the wrapper should indeed be a haskell function. (But as yet we still don't know how to generate code for wrappers...) M ./src/compiler98/Foreign.hs -9 +16

1 0

patch applied (hat): Even more 15-tuple stuff...
by Malcolm Wallace 10 Oct '06

10 Oct '06

Sun Jun 13 06:49:36 PDT 2004 panne * Even more 15-tuple stuff... M ./src/compiler98/TokenInt.hs -1 +1

1 0

patch applied (hat): Fixed prototype generation for functions returning a FunPtr, e.g.:
by Malcolm Wallace 10 Oct '06

10 Oct '06

Sat Jun 12 07:20:41 PDT 2004 panne * Fixed prototype generation for functions returning a FunPtr, e.g.: foreign import ccall unsafe "foo" foo :: CString -> IO (FunPtr a) The former prototype was: extern NodePtr(*)() foo(void*); but this is not really C, we want: extern NodePtr(*foo(void*))(); The fix could probably need some review & cleanup => Malcolm? :-) M ./src/compiler98/Foreign.hs +6

1 0