1. 29 Jun, 2000 1 commit
  2. 25 May, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-05-25 12:41:14 by simonpj] · 495ef8bd
      simonpj authored
      ~~~~~~~~~~~~
      		Apr/May 2000
      		~~~~~~~~~~~~
      
      This is a pretty big commit!  It adds stuff I've been working on
      over the last month or so.  DO NOT MERGE IT WITH 4.07!
      
      Interface file formats have changed a little; you'll need
      to make clean before remaking.
      
      						Simon PJ
      
      Recompilation checking
      ~~~~~~~~~~~~~~~~~~~~~~
      Substantial improvement in recompilation checking.  The version management
      is now entirely internal to GHC.  ghc-iface.lprl is dead!
      
      The trick is to generate the new interface file in two steps:
        - first convert Types etc to HsTypes etc, and thereby
      	build a new ParsedIface
        - then compare against the parsed (but not renamed) version of the old
      	interface file
      Doing this meant adding code to convert *to* HsSyn things, and to
      compare HsSyn things for equality.  That is the main tedious bit.
      
      Another improvement is that we now track version info for
      fixities and rules, which was missing before.
      
      
      Interface file reading
      ~~~~~~~~~~~~~~~~~~~~~~
      Make interface files reading more robust.
        * If the old interface file is unreadable, don't fail. [bug fix]
      
        * If the old interface file mentions interfaces
          that are unreadable, don't fail. [bug fix]
      
        * When we can't find the interface file,
          print the directories we are looking in.  [feature]
      
      
      Type signatures
      ~~~~~~~~~~~~~~~
        * New flag -ddump-types to print type signatures
      
      
      Type pruning
      ~~~~~~~~~~~~
      When importing
      	data T = T1 A | T2 B | T3 C
      it seems excessive to import the types A, B, C as well, unless
      the constructors T1, T2 etc are used.  A,B,C might be more types,
      and importing them may mean reading more interfaces, and so on.
       So the idea is that the renamer will just import the decl
      	data T
      unless one of the constructors is used.  This turns out to be quite
      easy to implement.  The downside is that we must make sure the
      constructors are always available if they are really needed, so
      I regard this as an experimental feature.
      
      
      Elimininate ThinAir names
      ~~~~~~~~~~~~~~~~~~~~~~~~~
      Eliminate ThinAir.lhs and all its works.  It was always a hack, and now
      the desugarer carries around an environment I think we can nuke ThinAir
      altogether.
      
      As part of this, I had to move all the Prelude RdrName defns from PrelInfo
      to PrelMods --- so I renamed PrelMods as PrelNames.
      
      I also had to move the builtinRules so that they are injected by the renamer
      (rather than appearing out of the blue in SimplCore).  This is if anything simpler.
      
      Miscellaneous
      ~~~~~~~~~~~~~
      * Tidy up the data types involved in Rules
      
      * Eliminate RnEnv.better_provenance; use Name.hasBetterProv instead
      
      * Add Unique.hasKey :: Uniquable a => a -> Unique -> Bool
        It's useful in a lot of places
      
      * Fix a bug in interface file parsing for __U[!]
      495ef8bd
  3. 24 May, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-05-24 15:47:13 by simonpj] · 95929be0
      simonpj authored
      MERGE 4.07
      
      * This fix cures the weird 'ifaceBinds' error that
        Sven and George tripped over.  It was quite obscure!
      
        Basically, there was a top level binding
      	f = x
        lying around, which CoreToStg didn't like.  Why hadn't
        it been substituted away?  Because it had a NOINLINE
        pragma.  Why did it have a NOINLINE pragma?  Because
        it's an always-diverging function, so we never want to
        inline it.
      95929be0
  4. 23 May, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-05-23 11:35:36 by simonpj] · bb91427f
      simonpj authored
      *** MERGE WITH 4.07 (once I've checked it works) ***
      
      * Fix result type signatures.  Note that a consequential change is that
        an ordinary binding with a variable on the left
      	f = e
        is now treated as a FunMonoBind, not a PatMonoBind.  This makes
        a few things a bit simpler (eg rnMethodBinds)
      
      * Fix warnings for unused imports.  This meant moving where provenances
        are improved in RnNames.  Move mkExportAvails from RnEnv to RnNames.
      
      * Print module names right (small change in Module.lhs and Rename.lhs)
      
      * Remove a few unused bindings
        
      * Add a little hack to let us print info about join points that turn
        out not to be let-no-escaped.  The idea is to call them "$j" and report
        any such variables that are not let-no-escaped.
      
      * Some small things aiming towards -ddump-types (harmless but incomplete)
      bb91427f
  5. 21 Apr, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-04-21 14:40:48 by simonpj] · 0666c3bb
      simonpj authored
      Remove a way to make the simplifier go into an
      infinite loop.   This has been there for some weeks;
      and George's UniForm tickled it.  I'm amazed nothing
      else has done so.  I'll add a test.
      
      The bad case happened when you go
      
      	let xs = 1:xs
      	in
      	foldr k z xs
      
      Then we kept firing the foldr/cons rule.
      
      Solution: we don't attach an unfolding *at all* to
      loop breakers (Simplify.completeBinding)
      0666c3bb
  6. 14 Apr, 2000 1 commit
  7. 13 Apr, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-04-13 11:56:35 by simonpj] · 9579283c
      simonpj authored
      Add support for 'packages'.
      
      * A package is a group of modules.
      
      * A package has a name (e.g. std)
      
      * A package is built into a single library (Unix; e.g. libHSstd.a)
                             or a single DLL     (Windows; e.g. HSstd.dll)
      
      * The '-package-name foo' flag tells GHC that the module being compiled
        is destined for package foo.
      
      * The '-package foo' flag tells GHC to make available modules
        from package 'foo'.  It replaces '-syslib foo' which is now deprecated.
      
      * Cross-package references cost an extra indirection in Windows,
        but not Unix
      
      * GHC does not maintain detailed cross-package dependency information.
        It does remember which modules in other packages the current module
        depends on, but not which things within those imported things.
      
      
      All of this tidies up the Prelude enormously.  The Prelude and
      Standard Libraries are built into a singl package called 'std'.  (This
      is a change; the library is now called libHSstd.a instead of libHS.a)
      9579283c
  8. 23 Mar, 2000 2 commits
    • simonpj's avatar
      [project @ 2000-03-23 17:45:17 by simonpj] · 111cee3f
      simonpj authored
      This utterly gigantic commit is what I've been up to in background
      mode in the last couple of months.  Originally the main goal
      was to get rid of Con (staturated constant applications)
      in the CoreExpr type, but one thing led to another, and I kept
      postponing actually committing.   Sorry.
      
      	Simon, 23 March 2000
      
      
      I've tested it pretty thoroughly, but doubtless things will break.
      
      Here are the highlights
      
      * Con is gone; the CoreExpr type is simpler
      * NoRepLits have gone
      * Better usage info in interface files => less recompilation
      * Result type signatures work
      * CCall primop is tidied up
      * Constant folding now done by Rules
      * Lots of hackery in the simplifier
      * Improvements in CPR and strictness analysis
      
      Many bug fixes including
      
      * Sergey's DoCon compiles OK; no loop in the strictness analyser
      * Volker Wysk's programs don't crash the CPR analyser
      
      I have not done much on measuring compilation times and binary sizes;
      they could have got worse.  I think performance has got significantly
      better, though, in most cases.
      
      
      Removing the Con form of Core expressions
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      The big thing is that
      
        For every constructor C there are now *two* Ids:
      
      	C is the constructor's *wrapper*. It evaluates and unboxes arguments
      	before calling $wC.  It has a perfectly ordinary top-level defn
      	in the module defining the data type.
      
      	$wC is the constructor's *worker*.  It is like a primop that simply
      	allocates and builds the constructor value.  Its arguments are the
      	actual representation arguments of the constructor.
      	Its type may be different to C, because:
      		- useless dict args are dropped
      		- strict args may be flattened
      
        For every primop P there is *one* Id, its (curried) Id
      
        Neither contructor worker Id nor the primop Id have a defminition anywhere.
        Instead they are saturated during the core-to-STG pass, and the code generator
        generates code for them directly. The STG language still has saturated
        primops and constructor applications.
      
      * The Const type disappears, along with Const.lhs.  The literal part
        of Const.lhs reappears as Literal.lhs.  Much tidying up in here,
        to bring all the range checking into this one module.
      
      * I got rid of NoRep literals entirely.  They just seem to be too much trouble.
      
      * Because Con's don't exist any more, the funny C { args } syntax
        disappears from inteface files.
      
      
      Parsing
      ~~~~~~~
      * Result type signatures now work
      	f :: Int -> Int = \x -> x
      	-- The Int->Int is the type of f
      
      	g x y :: Int = x+y
      	-- The Int is the type of the result of (g x y)
      
      
      Recompilation checking and make
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      * The .hi file for a modules is not touched if it doesn't change.  (It used to
        be touched regardless, forcing a chain of recompilations.)  The penalty for this
        is that we record exported things just as if they were mentioned in the body of
        the module.  And the penalty for that is that we may recompile a module when
        the only things that have changed are the things it is passing on without using.
        But it seems like a good trade.
      
      * -recomp is on by default
      
      Foreign declarations
      ~~~~~~~~~~~~~~~~~~~~
      * If you say
      	foreign export zoo :: Int -> IO Int
        then you get a C produre called 'zoo', not 'zzoo' as before.
        I've also added a check that complains if you export (or import) a C
        procedure whose name isn't legal C.
      
      
      Code generation and labels
      ~~~~~~~~~~~~~~~~~~~~~~~~~~
      * Now that constructor workers and wrappers have distinct names, there's
        no need to have a Foo_static_closure and a Foo_closure for constructor Foo.
        I nuked the entire StaticClosure story.  This has effects in some of
        the RTS headers (i.e. s/static_closure/closure/g)
      
      
      Rules, constant folding
      ~~~~~~~~~~~~~~~~~~~~~~~
      * Constant folding becomes just another rewrite rule, attached to the Id for the
        PrimOp.   To achieve this, there's a new form of Rule, a BuiltinRule (see CoreSyn.lhs).
        The prelude rules are in prelude/PrelRules.lhs, while simplCore/ConFold.lhs has gone.
      
      * Appending of constant strings now works, using fold/build fusion, plus
        the rewrite rule
      	unpack "foo" c (unpack "baz" c n)  =  unpack "foobaz" c n
        Implemented in PrelRules.lhs
      
      * The CCall primop is tidied up quite a bit.  There is now a data type CCall,
        defined in PrimOp, that packages up the info needed for a particular CCall.
        There is a new Id for each new ccall, with an big "occurrence name"
      	{__ccall "foo" gc Int# -> Int#}
        In interface files, this is parsed as a single Id, which is what it is, really.
      
      Miscellaneous
      ~~~~~~~~~~~~~
      * There were numerous places where the host compiler's
        minInt/maxInt was being used as the target machine's minInt/maxInt.
        I nuked all of these; everything is localised to inIntRange and inWordRange,
        in Literal.lhs
      
      * Desugaring record updates was broken: it didn't generate correct matches when
        used withe records with fancy unboxing etc.  It now uses matchWrapper.
      
      * Significant tidying up in codeGen/SMRep.lhs
      
      * Add __word, __word64, __int64 terminals to signal the obvious types
        in interface files.  Add the ability to print word values in hex into
        C code.
      
      * PrimOp.lhs is no longer part of a loop.  Remove PrimOp.hi-boot*
      
      
      Types
      ~~~~~
      * isProductTyCon no longer returns False for recursive products, nor
        for unboxed products; you have to test for these separately.
        There's no reason not to do CPR for recursive product types, for example.
        Ditto splitProductType_maybe.
      
      Simplification
      ~~~~~~~~~~~~~~~
      * New -fno-case-of-case flag for the simplifier.  We use this in the first run
        of the simplifier, where it helps to stop messing up expressions that
        the (subsequent) full laziness pass would otherwise find float out.
        It's much more effective than previous half-baked hacks in inlining.
      
        Actually, it turned out that there were three places in Simplify.lhs that
        needed to know use this flag.
      
      * Make the float-in pass push duplicatable bindings into the branches of
        a case expression, in the hope that we never have to allocate them.
        (see FloatIn.sepBindsByDropPoint)
      
      * Arrange that top-level bottoming Ids get a NOINLINE pragma
        This reduced gratuitous inlining of error messages.
        But arrange that such things still get w/w'd.
      
      * Arrange that a strict argument position is regarded as an 'interesting'
        context, so that if we see
      	foldr k z (g x)
        then we'll be inclined to inline g; this can expose a build.
      
      * There was a missing case in CoreUtils.exprEtaExpandArity that meant
        we were missing some obvious cases for eta expansion
        Also improve the code when handling applications.
      
      * Make record selectors (identifiable by their IdFlavour) into "cheap" operations.
      	  [The change is a 2-liner in CoreUtils.exprIsCheap]
        This means that record selection may be inlined into function bodies, which
        greatly improves the arities of overloaded functions.
      
      * Make a cleaner job of inlining "lone variables".  There was some distributed
        cunning, but I've centralised it all now in SimplUtils.analyseCont, which
        analyses the context of a call to decide whether it is "interesting".
      
      * Don't specialise very small functions in Specialise.specDefn
        It's better to inline it.  Rather like the worker/wrapper case.
      
      * Be just a little more aggressive when floating out of let rhss.
        See comments with Simplify.wantToExpose
        A small change with an occasional big effect.
      
      * Make the inline-size computation think that
      	case x of I# x -> ...
        is *free*.
      
      
      CPR analysis
      ~~~~~~~~~~~~
      * Fix what was essentially a bug in CPR analysis.  Consider
      
      	letrec f x = let g y = let ... in f e1
      		     in
      		     if ... then (a,b) else g x
      
        g has the CPR property if f does; so when generating the final annotated
        RHS for f, we must use an envt in which f is bound to its final abstract
        value.  This wasn't happening.  Instead, f was given the CPR tag but g
        wasn't; but of course the w/w pass gives rotten results in that case!!
        (Because f's CPR-ness relied on g's.)
      
        On they way I tidied up the code in CprAnalyse.  It's quite a bit shorter.
      
        The fact that some data constructors return a constructed product shows
        up in their CPR info (MkId.mkDataConId) not in CprAnalyse.lhs
      
      
      
      Strictness analysis and worker/wrapper
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      * BIG THING: pass in the demand to StrictAnal.saExpr.  This affects situations
        like
      	f (let x = e1 in (x,x))
        where f turns out to have strictness u(SS), say.  In this case we can
        mark x as demanded, and use a case expression for it.
      
        The situation before is that we didn't "know" that there is the u(SS)
        demand on the argument, so we simply computed that the body of the let
        expression is lazy in x, and marked x as lazily-demanded.  Then even after
        f was w/w'd we got
      
      	let x = e1 in case (x,x) of (a,b) -> $wf a b
      
        and hence
      
      	let x = e1 in $wf a b
      
        I found a much more complicated situation in spectral/sphere/Main.shade,
        which improved quite a bit with this change.
      
      * Moved the StrictnessInfo type from IdInfo to Demand.  It's the logical
        place for it, and helps avoid module loops
      
      * Do worker/wrapper for coerces even if the arity is zero.  Thus:
      	stdout = coerce Handle (..blurg..)
        ==>
      	wibble = (...blurg...)
      	stdout = coerce Handle wibble
        This is good because I found places where we were saying
      	case coerce t stdout of { MVar a ->
      	...
      	case coerce t stdout of { MVar b ->
      	...
        and the redundant case wasn't getting eliminated because of the coerce.
      111cee3f
    • sewardj's avatar
      [project @ 2000-03-23 12:22:04 by sewardj] · 1e5271f1
      sewardj authored
      In interface files, don't forget to mention the names of modules imported
      via hi-boot files.  This is needed so that Hugs can use the import decls
      in interface files to safely overestimate the dependency sets which it will
      encounter when linking object code.
      1e5271f1
  9. 02 Mar, 2000 1 commit
  10. 25 Feb, 2000 1 commit
    • panne's avatar
      [project @ 2000-02-25 14:55:31 by panne] · f83a5a68
      panne authored
      Deprecations of variables now works, although the source location is
      not yet reported correctly and the code needs some cleanup. Added a
      new flag -fwarn-deprecations to the set of standard warnings. The
      syntax of deprecations has been extended to deprecate types, classes,
      or even constructors, although this does not work yet.
      f83a5a68
  11. 22 Feb, 2000 1 commit
  12. 21 Feb, 2000 1 commit
  13. 20 Feb, 2000 1 commit
  14. 09 Feb, 2000 1 commit
  15. 28 Jan, 2000 1 commit
    • lewie's avatar
      [project @ 2000-01-28 20:52:37 by lewie] · 266fadd9
      lewie authored
      First pass at implicit parameters.  Honest, I didn't really go in *intending*
      to modify every file in the typechecker... ;-)  The breadth of the change
      is partly due to generalizing contexts so that they are not hardwired to
      be (Class, [Type]) pairs.  See types/Type.lhs for details (look for PredType).
      266fadd9
  16. 04 Jan, 2000 1 commit
    • simonpj's avatar
      [project @ 2000-01-04 17:40:46 by simonpj] · 5ca77490
      simonpj authored
      This commit arranges that literal strings will fuse
      nicely, by expressing them as an application of build.
      
      * NoRepStr is now completely redundant, though I havn't removed it yet.
      
      * The unpackStr stuff moves from PrelPack to PrelBase.
      
      * There's a new form of Rule, a BuiltinRule, for rules that
        can't be expressed in Haskell.  The string-fusion rule is one
        such.  It's defined in prelude/PrelRules.lhs.
      
      * PrelRules.lhs also contains a great deal of code that
        implements constant folding.  In due course this will replace
        ConFold.lhs, but for the moment it simply duplicates it.
      5ca77490
  17. 20 Dec, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-12-20 10:34:27 by simonpj] · e921b2e3
      simonpj authored
      This commit implements a substantial re-organisation of the Prelude
      It also fixes a couple of small renamer bugs that were reported recently
      	(notably, Sven pointed out that we weren't reporting
      	unused imports properly)
      
      My original goal was to get rid of all "orphan" modules (i.e. ones
      with instance decls that don't belong either to a tycon or a class
      defined in the same module).  This should reduce the number of
      interface files that have to be read when compiling small Haskell
      modules.
      
      But like most expeditions into the Prelude Swamp, it spiraled out
      of control.  The result is quite satisfactory, though.
      
      	GONE AWAY:	PrelCCall, PrelNumExtra
      
      	NEW:		PrelReal, PrelFloat, PrelByteArr, PrelNum.hi-boot
      
      (The extra PrelNum.hi-boot is because of a tiresome thin-air Id, addr2Integer,
      which used to be in PrelBase.)
      
      Quite a lot of types have moved from one module to another,
      which entails some changes to part of the compiler (PrelInfo, PrelMods) etc,
      and there are a few places in the RTS includes and even in the driver
      that know about these home modules (alas).
      
      So the rough structure is as follows, in (linearised) dependency order
      	[this list now appears in PrelBase.lhs]
      
      PrelGHC		Has no implementation.  It defines built-in things, and
      		by importing it you bring them into scope.
      		The source file is PrelGHC.hi-boot, which is just
      		copied to make PrelGHC.hi
      
      		Classes: CCallable, CReturnable
      
      PrelBase	Classes: Eq, Ord, Functor, Monad
      		Types:   list, (), Int, Bool, Ordering, Char, String
      
      PrelTup		Types: tuples, plus instances for PrelBase classes
      
      PrelShow	Class: Show, plus instances for PrelBase/PrelTup types
      
      PrelEnum	Class: Enum,  plus instances for PrelBase/PrelTup types
      
      PrelMaybe	Type: Maybe, plus instances for PrelBase classes
      
      PrelNum		Class: Num, plus instances for Int
      		Type:  Integer, plus instances for all classes so far (Eq, Ord, Num, Show)
      
      		Integer is needed here because it is mentioned in the signature
      		of 'fromInteger' in class Num
      
      PrelReal	Classes: Real, Integral, Fractional, RealFrac
      			 plus instances for Int, Integer
      		Types:  Ratio, Rational
      			plus intances for classes so far
      
      		Rational is needed here because it is mentioned in the signature
      		of 'toRational' in class Real
      
      Ix		Classes: Ix, plus instances for Int, Bool, Char, Integer, Ordering, tuples
      
      PrelArr		Types: Array, MutableArray, MutableVar
      
      		Does *not* contain any ByteArray stuff (see PrelByteArr)
      		Arrays are used by a function in PrelFloat
      
      PrelFloat	Classes: Floating, RealFloat
      		Types:   Float, Double, plus instances of all classes so far
      
      		This module contains everything to do with floating point.
      		It is a big module (900 lines)
      		With a bit of luck, many modules can be compiled without ever reading PrelFloat.hi
      
      PrelByteArr	Types: ByteArray, MutableByteArray
      
      		We want this one to be after PrelFloat, because it defines arrays
      		of unboxed floats.
      
      
      Other Prelude modules are much easier with fewer complex dependencies.
      e921b2e3
  18. 09 Dec, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-12-09 12:30:56 by simonpj] · 39262efa
      simonpj authored
      A bunch of small changes in the way that usage information
      is generated to stuff into interface files.  I'm not certain
      anything was really wrong before, but it's tidier now, and
      there are slightly fewer dependencies generated.  Main
      differences are in RnIfaces.getImportVersions
      
      I also made the check for orphan rules a bit cleverer
      (Rename.isOrphanDecl) so that we get fewer spurious orphan modules.
      
      Simon
      39262efa
  19. 30 Nov, 1999 1 commit
    • lewie's avatar
      [project @ 1999-11-30 16:10:07 by lewie] · e1e1d020
      lewie authored
      First bits o' functional dependencies - just the syntax and related
      datatypes, plus started moving some of the static checks from the
      renamer (where we don't know about fundeps) to later in the typechecker.
      e1e1d020
  20. 29 Nov, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-11-29 17:34:14 by simonpj] · e1a4f2a5
      simonpj authored
      Make it so that a class decl generates default method decls
      for every method, not just for the ones that the user supplies
      default-methods for.
      
      GHC will never call these default-default methods, because
      when it finds an instance decl with no defn for a method,
      *and* the class decl doesn't have a user-programmed default
      method, it whips up a new default method for that instance
      decl so that the error message is more informative than
      the default-default method would be.
      
      But Hugs isn't so smart, and wants to call something from
      the class decl.
      
      This change required fiddling with more than I expected.  Sigh.
      
      Simon
      e1a4f2a5
  21. 01 Nov, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-11-01 17:09:54 by simonpj] · 30b5ebe4
      simonpj authored
      A regrettably-gigantic commit that puts in place what Simon PJ
      has been up to for the last month or so, on and off.
      
      The basic idea was to restore unfoldings to *occurrences* of
      variables without introducing a space leak.  I wanted to make
      sure things improved relative to 4.04, and that proved depressingly
      hard.  On the way I discovered several quite serious bugs in the
      simplifier.
      
      Here's a summary of what's gone on.
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      * No commas between for-alls in RULES.  This makes the for-alls have
        the same syntax as in types.
      
      * Arrange that simplConArgs works in one less pass than before.
        This exposed a bug: a bogus call to completeBeta.
      
      * Add a top-level flag in CoreUnfolding, used in callSiteInline
      
      * Extend w/w to use etaExpandArity, so it does eta/coerce expansion
      
      * Implement inline phases.   The meaning of the inline pragmas is
        described in CoreUnfold.lhs.  You can say things like
      	{#- INLINE 2 build #-}
        to mean "inline build in phase 2"
      
      * Don't float anything out of an INLINE.
        Don't float things to top level unless they also escape a value lambda.
      	[see comments with SetLevels.lvlMFE
        Without at least one of these changes, I found that
      	{-# INLINE concat #-}
      	concat = __inline (/\a -> foldr (++) [])
        was getting floated to
      	concat = __inline( /\a -> lvl a )
      	lvl = ...inlined version of foldr...
      
        Subsequently I found that not floating constants out of an INLINE
        gave really bad code like
      	__inline (let x = e in \y -> ...)
        so I now let things float out of INLINE
      
      * Implement the "reverse-mapping" idea for CSE; actually it turned out to be easier
        to implement it in SetLevels, and may benefit full laziness too.
      
      * It's a good idea to inline inRange. Consider
      
      	index (l,h) i = case inRange (l,h) i of
      		  	  True ->  l+i
      			  False -> error
        inRange itself isn't strict in h, but if it't inlined then 'index'
        *does* become strict in h.  Interesting!
      
      * Big change to the way unfoldings and occurrence info is propagated in the simplifier
        The plan is described in Subst.lhs with the Subst type
        Occurrence info is now in a separate IdInfo field than user pragmas
      
      * I found that
      	(coerce T (coerce S (\x.e))) y
        didn't simplify in one round. First we get to
      	(\x.e) y
        and only then do the beta. Solution: cancel the coerces in the continuation
      
      * Amazingly, CoreUnfold wasn't counting the cost of a function an application.
      
      * Disable rules in initial simplifier run.  Otherwise full laziness
        doesn't get a chance to lift out a MFE before a rule (e.g. fusion)
        zaps it.  queens is a case in point
      
      * Improve float-out stuff significantly.  The big change is that if we have
      
      	\x -> ... /\a -> ...let p = ..a.. in let q = ...p...
      
        where p's rhs doesn't x, we abstract a from p, so that we can get p past x.
        (We did that before.)  But we also substitute (p a) for p in q, and then
        we can do the same thing for q.  (We didn't do that, so q got stuck.)
        This is much better.  It involves doing a substitution "as we go" in SetLevels,
        though.
      30b5ebe4
  22. 17 Sep, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-09-17 09:15:22 by simonpj] · 731f53de
      simonpj authored
      This bunch of commits represents work in progress on inlining and
      worker/wrapper stuff.
      
      Currently, I think it makes the compiler slightly worse than 4.04, for
      reasons I don't yet understand.  But it means that Simon and I can
      both peer at what is going on.
      
      * Substantially improve handling of coerces in worker/wrapper
      
      * exprIsDupable for an application (f e1 .. en) wasn't calling exprIsDupable
        on the arguments!!  So applications with few, but large, args were being dupliated.
      
      * sizeExpr on an application wasn't doing a nukeScrutDiscount on the arg of
        an application!!  So bogus discounts could accumulate from arguments!
      
      * Improve handling of INLINE pragmas in calcUnfoldingGuidance.  It was really
        wrong before
      731f53de
  23. 15 Jul, 1999 1 commit
    • keithw's avatar
      [project @ 1999-07-15 14:08:03 by keithw] · 9d787ef5
      keithw authored
      This commit makes a start at implementing polymorphic usage
      annotations.
      
      * The module Type has now been split into TypeRep, containing the
        representation Type(..) and other information for `friends' only,
        and Type, providing the public interface to Type.  Due to a bug in
        the interface-file slurping prior to ghc-4.04, {-# SOURCE #-}
        dependencies must unfortunately still refer to TypeRep even though
        they are not friends.
      
      * Unfoldings in interface files now print as __U instead of __u.
        UpdateInfo now prints as __UA instead of __U.
      
      * A new sort of variables, UVar, in their own namespace, uvName, has
        been introduced for usage variables.
      
      * Usage binders __fuall uv have been introduced.  Usage annotations
        are now __u - ty (used once), __u ! ty (used possibly many times),
        __u uv ty (used uv times), where uv is a UVar.  __o and __m have
        gone.  All this still lives only in a TyNote, *for now* (but not for
        much longer).
      
      * Variance calculation for TyCons has moved from
        typecheck/TcTyClsDecls to types/Variance.
      
      * Usage annotation and inference are now done together in a single
        pass.  Provision has been made for inferring polymorphic usage
        annotations (with __fuall) but this has not yet been implemented.
        Watch this space!
      9d787ef5
  24. 14 Jul, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-07-14 14:40:20 by simonpj] · 4e7d56fd
      simonpj authored
      Main things:
      
      * Add splitProductType_maybe to DataCon.lhs, with type
        splitProductType_maybe
      	:: Type 			-- A product type, perhaps
      	-> Maybe (TyCon, 		-- The type constructor
      		  [Type],		-- Type args of the tycon
      		  DataCon,		-- The data constructor
      		  [Type])		-- Its *representation* arg types
      
        Then use it in many places (e.g. worker-wrapper places) instead
        of a pile of junk
      
      * Clean up various uses of dataConArgTys, which were plain wrong because
        they weren't passed the existential type arguments.  Most of these calls
        are eliminated by using splitProductType_maybe above.  I hope I correctly
        squashed the others. This fixes a bug that Meurig's programs showed up.
      
          module FailGHC (killSustainer) where
          import Weak
          import IOExts
      
          data Sustainer = forall a . Sustainer (IORef (Maybe a)) (IO ())
      
          killSustainer :: Sustainer -> IO ()
          killSustainer (Sustainer _ act) = act
      
        The above program used to kill the compiler.
      
      * A fairly concerted attack on the Dreaded Space Leak.
      	- Add Type.seqType, CoreSyn.seqExpr, CoreSyn.seqRules
      
      	- Add some seq'ing when building Ids and IdInfos
      		These reduce the space usage a lot
      
      	- Add CoreSyn.coreBindsSize, which is pretty strict in the program,
      		and call it when we have -dshow-passes.
      
      	- Do not put the inlining in an Id that is being plugged into
      		the result-expression of the simplifier.  This cures
      		a the 'wedge' in the space profile for reasons I don't understand fully
      
        Together, these things reduce the max space usage when compiling PrelNum from
        17M to about 7Mbytes.
      
        I think there are now *too many* seqs, and they waste work, but I don't have
        time to find which ones.
      
        Furthermore, we aren't done. For some reason, some of the stuff allocated by
        the simplifier makes it through all during code generation and I don't see why.
        There's a should-be-unnecessary call to coreBindsSize in Main.main which
        zaps some, but not all of this space.
      
        -dshow-passes reduces space usage a bit, but I don't think it should really.
      
        All the measurements were made on a compiler compiled with profiling by
        GHC 3.03.    I hope they carry over to other builds!
      
      * One trivial thing: changed all variables 'label' to 'lbl', becuase the
        former is a keyword with -fglagow-exts in GHC 3.03 (which I was compiling with).
        Something similar in StringBuffer.
      4e7d56fd
  25. 06 Jul, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-07-06 16:45:31 by simonpj] · 9d38678e
      simonpj authored
      All Simon's recent tuning changes.  Rough summary follows:
      
      * Fix Kevin Atkinson's cant-find-instance bug.  Turns out that Rename.slurpSourceRefs
        needs to repeatedly call getImportedInstDecls, and then go back to slurping
        source-refs.  Comments with Rename.slurpSourceRefs.
      
      * Add a case to Simplify.mkDupableAlt for the quite-common case where there's
        a very simple alternative, in which case there's no point in creating a
        join-point binding.
      
      * Fix CoreUtils.exprOkForSpeculation so that it returns True of (==# a# b#).
        This lack meant that
      	case ==# a# b# of { True -> x; False -> x }
        was not simplifying
      
      * Make float-out dump bindings at the top of a function argument, as
        at the top of a let(rec) rhs.  See notes with FloatOut.floatRhs
      
      * Make the ArgOf case of mkDupableAlt generate a OneShot lambda.
        This gave a noticeable boost to spectral/boyer2
      
      
      * Reduce the number of coerces, using worker/wrapper stuff.
        The main idea is in WwLib.mkWWcoerce.  The gloss is that we must do
        the w/w split even for small non-recursive things.  See notes with
        WorkWrap.tryWw.
      
      * This further complicated getWorkerId, so I finally bit the bullet and
        make the workerInfo field of the IdInfo work properly, including
        under substitutions.  Death to getWorkerId.  Kevin Glynn will be happy.
      
      * Make all lambdas over realWorldStatePrimTy
        into one-shot lambdas.  This is a GROSS HACK.
      
      * Also make the occurrence analyser aware of one-shot lambdas.
      
      * Make various Prelude things into INLINE, so that foldr doesn't
        get inlined in their body, so that the caller gets the benefit
        of fusion.  Notably in PrelArr.lhs.
      9d38678e
  26. 18 Jun, 1999 1 commit
  27. 18 May, 1999 1 commit
  28. 14 Apr, 1999 1 commit
    • kglynn's avatar
      [project @ 1999-04-14 04:07:57 by kglynn] · 27ce7270
      kglynn authored
      (keving)
      
      Sigh,  No point w/w pass storing the worker id in wrapper's info
      field,  because later simplifications may change the worker id. I've
      restored the original code which looks in the wrapper function and
      guesses the worker id.  With the addition of CPR transf. this looks a
      bit fragile. thinks, thinks, ....
      27ce7270
  29. 13 Apr, 1999 1 commit
    • kglynn's avatar
      [project @ 1999-04-13 08:55:33 by kglynn] · 30b5b5cc
      kglynn authored
      (keving)
      
      Big Bang introduction of CPR Analysis Pass.  Note that now
      -fstrictness only does the strictness analysis phase,  it is necessary
      to follow this with -fworker-wrapper to actually do the required Core
      transformations. The -O option in the ghc driver script has been
      modified appropriately.
      
      For now,  CPR analysis is turned off.  To try it,  insert a
      -fcpr_analyse between the -fstrictness and the -fworker-wrapper
      options.
      
      Misc. comments:
      
      - The worker flag has been removed from an ID's StrictnessInfo field.
      Now the worker info is an extra field in the Id's prag info.
      
      - We do a nested CPR analysis,  but worker-wrapper only looks at the
      info for the outermost constructor,  else laziness can be lost.
      
      - Id's CPR Info in traces and interfaces file follows __M
      
      - Worker-wrappery transformation now accounts for both strictness and
      CPR analysis results.
      30b5b5cc
  30. 06 Apr, 1999 1 commit
  31. 02 Mar, 1999 1 commit
  32. 28 Jan, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-01-28 09:19:57 by simonpj] · b70e2f94
      simonpj authored
      Always inline nullary constructors.  This makes a 
      difference in:
      
      	case x ># y of r {
      	  True  -> f1 r
      	  False -> f2 r
       	}
      
      The code generator currently has difficulty binding "r"
      to the boolean result of the comparision (and the compiler
      crashes).  This fix substitutes for r, thus:
      
      	case x ># y of r {
      	  True  -> f1 True
      	  False -> f2 False
       	}
      
      Voila.
      b70e2f94
  33. 27 Jan, 1999 1 commit
    • simonpj's avatar
      [project @ 1999-01-27 14:51:14 by simonpj] · 18976e61
      simonpj authored
      Finally!  This commits the ongoing saga of Simon's hygiene sweep
      
      FUNCTIONALITY
      ~~~~~~~~~~~~~
      a) The 'unused variable' warnings from the renamer work.  
      b) Better error messages here and there, esp type checker
      c) Fixities for Haskell 98 (maybe I'd done that before)
      d) Lazy reporting of name clashes for Haskell 98 (ditto)
      
      HYGIENE
      ~~~~~~~
      a) type OccName has its own module.  OccNames are represented
         by a single FastString, not three as in the last round.  This
         string is held in Z-encoded form; a decoding function decodes
         for printing in user error messages.  There's a nice tight
         encoding for (,,,,,,,,,,,,,,,,,,,,,,,,,,,,,)
      
      b) type Module is a proper ADT, in module OccName
      
      c) type RdrName is a proper ADT, in its own module
      
      d) type Name has a new, somwhat tidier, representation
      
      e) much grunting in the renamer to get Provenances right.
         This makes error messages look better (no spurious qualifiers)
      18976e61
  34. 22 Dec, 1998 1 commit
    • simonpj's avatar
      [project @ 1998-12-22 16:31:28 by simonpj] · 8295d9ca
      simonpj authored
      1.  Add primOpStrictness to PrimOp.lhs, and use it in
      	- the strictness analyser
      	- the simplifier
          to deal correctly with PrimOps that are non-strict.
      
         ToDo: use this new facility to clean up SeqOp, ParOp.
      
      2. Fix the instance-decl-import bug, but printing de-synonym'd types
         in interface files.
      
      3. Make the simplifier treat applications with an unlifted-type arg
         in the same way it would if the function was strict
         (in rebuild_strict)
      8295d9ca
  35. 18 Dec, 1998 1 commit
    • simonpj's avatar
      [project @ 1998-12-18 17:40:31 by simonpj] · 7e602b0a
      simonpj authored
      Another big commit from Simon.  Actually, the last one
      didn't all go into the main trunk; because of a CVS glitch it
      ended up in the wrong branch.
      
      So this commit includes:
      
      * Scoped type variables
      * Warnings for unused variables should work now (they didn't before)
      * Simplifier improvements:
      	- Much better treatment of strict arguments
      	- Better treatment of bottoming Ids
      	- No need for w/w split for fns that are merely strict
      	- Fewer iterations needed, I hope
      * Less gratuitous renaming in interface files and abs C
      * OccName is a separate module, and is an abstract data type
      
      I think the whole Prelude and Exts libraries compile correctly.
      Something isn't quite right about typechecking existentials though.
      7e602b0a
  36. 02 Dec, 1998 1 commit
  37. 14 Aug, 1998 1 commit
  38. 16 Jun, 1998 1 commit
    • sof's avatar
      [project @ 1998-06-16 12:25:36 by sof] · 1ca8c2c7
      sof authored
      New function: okToUnfoldInHiFile.
      
        - For values whose RHS have been deemed to be interface
          file unfolding candidates, do a last minute check to
          see whether the unfolding is really suitable.
      
          An unfolding is not suitable iff
      
             - contains a _casm_
             - contains a lit-lit
      
          The reason for not unfolding _casm_/lit-lits into
          interface files is that their C fragments are likely
          to mention #defines/functions that will be
          out-of-scope at an unfolding site. Turning off unfolding
          of such expressions avoid this unfortunate situation.
      1ca8c2c7
  39. 22 May, 1998 1 commit
    • simonm's avatar
      [project @ 1998-05-22 15:23:11 by simonm] · f36fb2ce
      simonm authored
      Add NOINLINE pragma.
      
      - add new type of inline info: IDontWantToBeINLINEd
      
      - hopefully get the interactions between IMustNotBeINLINEd (which is
        used by the simplifier to ensure termination when simplifying
        recursive binding groups) and IDontWantToBeINLINEd.
      
      - no need to pass NOINLINE across modules, we just make sure that any
        function marked as NOLINE doesn't get an unfolding in the interface.
      f36fb2ce