1 \section[GHC.Base]{Module @GHC.Base@}
3 The overall structure of the GHC Prelude is a bit tricky.
5 a) We want to avoid "orphan modules", i.e. ones with instance
6 decls that don't belong either to a tycon or a class
7 defined in the same module
9 b) We want to avoid giant modules
11 So the rough structure is as follows, in (linearised) dependency order
14 GHC.Prim Has no implementation. It defines built-in things, and
15 by importing it you bring them into scope.
16 The source file is GHC.Prim.hi-boot, which is just
17 copied to make GHC.Prim.hi
19 GHC.Base Classes: Eq, Ord, Functor, Monad
20 Types: list, (), Int, Bool, Ordering, Char, String
22 Data.Tuple Types: tuples, plus instances for GHC.Base classes
24 GHC.Show Class: Show, plus instances for GHC.Base/GHC.Tup types
26 GHC.Enum Class: Enum, plus instances for GHC.Base/GHC.Tup types
28 Data.Maybe Type: Maybe, plus instances for GHC.Base classes
30 GHC.List List functions
32 GHC.Num Class: Num, plus instances for Int
33 Type: Integer, plus instances for all classes so far (Eq, Ord, Num, Show)
35 Integer is needed here because it is mentioned in the signature
36 of 'fromInteger' in class Num
38 GHC.Real Classes: Real, Integral, Fractional, RealFrac
39 plus instances for Int, Integer
40 Types: Ratio, Rational
41 plus intances for classes so far
43 Rational is needed here because it is mentioned in the signature
44 of 'toRational' in class Real
46 GHC.ST The ST monad, instances and a few helper functions
48 Ix Classes: Ix, plus instances for Int, Bool, Char, Integer, Ordering, tuples
50 GHC.Arr Types: Array, MutableArray, MutableVar
52 Arrays are used by a function in GHC.Float
54 GHC.Float Classes: Floating, RealFloat
55 Types: Float, Double, plus instances of all classes so far
57 This module contains everything to do with floating point.
58 It is a big module (900 lines)
59 With a bit of luck, many modules can be compiled without ever reading GHC.Float.hi
62 Other Prelude modules are much easier with fewer complex dependencies.
71 , ExistentialQuantification
74 -- -fno-warn-orphans is needed for things like:
75 -- Orphan rule: "x# -# x#" ALWAYS forall x# :: Int# -# x# x# = 0
76 {-# OPTIONS_GHC -fno-warn-orphans #-}
77 {-# OPTIONS_HADDOCK hide #-}
79 -----------------------------------------------------------------------------
82 -- Copyright : (c) The University of Glasgow, 1992-2002
83 -- License : see libraries/base/LICENSE
85 -- Maintainer : cvs-ghc@haskell.org
86 -- Stability : internal
87 -- Portability : non-portable (GHC extensions)
89 -- Basic data types and classes.
91 -----------------------------------------------------------------------------
100 --module GHC.Generics, -- JPM: We no longer export GHC.Generics
101 -- by default to avoid name clashes
104 module GHC.Prim, -- Re-export GHC.Prim and GHC.Err, to avoid lots
105 module GHC.Err -- of people having to import it explicitly
111 -- JPM: Since we don't export it, we don't need to import GHC.Generics
112 --import GHC.Generics
115 import {-# SOURCE #-} GHC.Show
116 import {-# SOURCE #-} GHC.Err
117 import {-# SOURCE #-} GHC.IO (failIO)
119 -- These two are not strictly speaking required by this module, but they are
120 -- implicit dependencies whenever () or tuples are mentioned, so adding them
121 -- as imports here helps to get the dependencies right in the new build system.
131 default () -- Double isn't available yet
135 %*********************************************************
137 \subsection{DEBUGGING STUFF}
138 %* (for use when compiling GHC.Base itself doesn't work)
140 %*********************************************************
144 data Bool = False | True
145 data Ordering = LT | EQ | GT
153 (&&) True True = True
159 unpackCString# :: Addr# -> [Char]
160 unpackFoldrCString# :: Addr# -> (Char -> a -> a) -> a -> a
161 unpackAppendCString# :: Addr# -> [Char] -> [Char]
162 unpackCStringUtf8# :: Addr# -> [Char]
163 unpackCString# a = error "urk"
164 unpackFoldrCString# a = error "urk"
165 unpackAppendCString# a = error "urk"
166 unpackCStringUtf8# a = error "urk"
171 %*********************************************************
173 \subsection{Monadic classes @Functor@, @Monad@ }
175 %*********************************************************
178 {- | The 'Functor' class is used for types that can be mapped over.
179 Instances of 'Functor' should satisfy the following laws:
182 > fmap (f . g) == fmap f . fmap g
184 The instances of 'Functor' for lists, 'Data.Maybe.Maybe' and 'System.IO.IO'
188 class Functor f where
189 fmap :: (a -> b) -> f a -> f b
191 -- | Replace all locations in the input with the same value.
192 -- The default definition is @'fmap' . 'const'@, but this may be
193 -- overridden with a more efficient version.
194 (<$) :: a -> f b -> f a
197 {- | The 'Monad' class defines the basic operations over a /monad/,
198 a concept from a branch of mathematics known as /category theory/.
199 From the perspective of a Haskell programmer, however, it is best to
200 think of a monad as an /abstract datatype/ of actions.
201 Haskell's @do@ expressions provide a convenient syntax for writing
204 Minimal complete definition: '>>=' and 'return'.
206 Instances of 'Monad' should satisfy the following laws:
208 > return a >>= k == k a
210 > m >>= (\x -> k x >>= h) == (m >>= k) >>= h
212 Instances of both 'Monad' and 'Functor' should additionally satisfy the law:
214 > fmap f xs == xs >>= return . f
216 The instances of 'Monad' for lists, 'Data.Maybe.Maybe' and 'System.IO.IO'
217 defined in the "Prelude" satisfy these laws.
221 -- | Sequentially compose two actions, passing any value produced
222 -- by the first as an argument to the second.
223 (>>=) :: forall a b. m a -> (a -> m b) -> m b
224 -- | Sequentially compose two actions, discarding any value produced
225 -- by the first, like sequencing operators (such as the semicolon)
226 -- in imperative languages.
227 (>>) :: forall a b. m a -> m b -> m b
228 -- Explicit for-alls so that we know what order to
229 -- give type arguments when desugaring
231 -- | Inject a value into the monadic type.
233 -- | Fail with a message. This operation is not part of the
234 -- mathematical definition of a monad, but is invoked on pattern-match
235 -- failure in a @do@ expression.
236 fail :: String -> m a
239 m >> k = m >>= \_ -> k
244 %*********************************************************
246 \subsection{The list type}
248 %*********************************************************
251 instance Functor [] where
254 instance Monad [] where
255 m >>= k = foldr ((++) . k) [] m
256 m >> k = foldr ((++) . (\ _ -> k)) [] m
261 A few list functions that appear here because they are used here.
262 The rest of the prelude list functions are in GHC.List.
264 ----------------------------------------------
265 -- foldr/build/augment
266 ----------------------------------------------
269 -- | 'foldr', applied to a binary operator, a starting value (typically
270 -- the right-identity of the operator), and a list, reduces the list
271 -- using the binary operator, from right to left:
273 -- > foldr f z [x1, x2, ..., xn] == x1 `f` (x2 `f` ... (xn `f` z)...)
275 foldr :: (a -> b -> b) -> b -> [a] -> b
277 -- foldr f z (x:xs) = f x (foldr f z xs)
278 {-# INLINE [0] foldr #-}
279 -- Inline only in the final stage, after the foldr/cons rule has had a chance
280 -- Also note that we inline it when it has *two* parameters, which are the
281 -- ones we are keen about specialising!
285 go (y:ys) = y `k` go ys
287 -- | A list producer that can be fused with 'foldr'.
288 -- This function is merely
290 -- > build g = g (:) []
292 -- but GHC's simplifier will transform an expression of the form
293 -- @'foldr' k z ('build' g)@, which may arise after inlining, to @g k z@,
294 -- which avoids producing an intermediate list.
296 build :: forall a. (forall b. (a -> b -> b) -> b -> b) -> [a]
297 {-# INLINE [1] build #-}
298 -- The INLINE is important, even though build is tiny,
299 -- because it prevents [] getting inlined in the version that
300 -- appears in the interface file. If [] *is* inlined, it
301 -- won't match with [] appearing in rules in an importing module.
303 -- The "1" says to inline in phase 1
307 -- | A list producer that can be fused with 'foldr'.
308 -- This function is merely
310 -- > augment g xs = g (:) xs
312 -- but GHC's simplifier will transform an expression of the form
313 -- @'foldr' k z ('augment' g xs)@, which may arise after inlining, to
314 -- @g k ('foldr' k z xs)@, which avoids producing an intermediate list.
316 augment :: forall a. (forall b. (a->b->b) -> b -> b) -> [a] -> [a]
317 {-# INLINE [1] augment #-}
318 augment g xs = g (:) xs
321 "fold/build" forall k z (g::forall b. (a->b->b) -> b -> b) .
322 foldr k z (build g) = g k z
324 "foldr/augment" forall k z xs (g::forall b. (a->b->b) -> b -> b) .
325 foldr k z (augment g xs) = g k (foldr k z xs)
327 "foldr/id" foldr (:) [] = \x -> x
328 "foldr/app" [1] forall ys. foldr (:) ys = \xs -> xs ++ ys
329 -- Only activate this from phase 1, because that's
330 -- when we disable the rule that expands (++) into foldr
332 -- The foldr/cons rule looks nice, but it can give disastrously
333 -- bloated code when commpiling
334 -- array (a,b) [(1,2), (2,2), (3,2), ...very long list... ]
335 -- i.e. when there are very very long literal lists
336 -- So I've disabled it for now. We could have special cases
337 -- for short lists, I suppose.
338 -- "foldr/cons" forall k z x xs. foldr k z (x:xs) = k x (foldr k z xs)
340 "foldr/single" forall k z x. foldr k z [x] = k x z
341 "foldr/nil" forall k z. foldr k z [] = z
343 "augment/build" forall (g::forall b. (a->b->b) -> b -> b)
344 (h::forall b. (a->b->b) -> b -> b) .
345 augment g (build h) = build (\c n -> g c (h c n))
346 "augment/nil" forall (g::forall b. (a->b->b) -> b -> b) .
347 augment g [] = build g
350 -- This rule is true, but not (I think) useful:
351 -- augment g (augment h t) = augment (\cn -> g c (h c n)) t
355 ----------------------------------------------
357 ----------------------------------------------
360 -- | 'map' @f xs@ is the list obtained by applying @f@ to each element
363 -- > map f [x1, x2, ..., xn] == [f x1, f x2, ..., f xn]
364 -- > map f [x1, x2, ...] == [f x1, f x2, ...]
366 map :: (a -> b) -> [a] -> [b]
368 map f (x:xs) = f x : map f xs
371 mapFB :: (elt -> lst -> lst) -> (a -> elt) -> a -> lst -> lst
372 {-# INLINE [0] mapFB #-}
373 mapFB c f = \x ys -> c (f x) ys
375 -- The rules for map work like this.
377 -- Up to (but not including) phase 1, we use the "map" rule to
378 -- rewrite all saturated applications of map with its build/fold
379 -- form, hoping for fusion to happen.
380 -- In phase 1 and 0, we switch off that rule, inline build, and
381 -- switch on the "mapList" rule, which rewrites the foldr/mapFB
382 -- thing back into plain map.
384 -- It's important that these two rules aren't both active at once
385 -- (along with build's unfolding) else we'd get an infinite loop
386 -- in the rules. Hence the activation control below.
388 -- The "mapFB" rule optimises compositions of map.
390 -- This same pattern is followed by many other functions:
391 -- e.g. append, filter, iterate, repeat, etc.
394 "map" [~1] forall f xs. map f xs = build (\c n -> foldr (mapFB c f) n xs)
395 "mapList" [1] forall f. foldr (mapFB (:) f) [] = map f
396 "mapFB" forall c f g. mapFB (mapFB c f) g = mapFB c (f.g)
401 ----------------------------------------------
403 ----------------------------------------------
405 -- | Append two lists, i.e.,
407 -- > [x1, ..., xm] ++ [y1, ..., yn] == [x1, ..., xm, y1, ..., yn]
408 -- > [x1, ..., xm] ++ [y1, ...] == [x1, ..., xm, y1, ...]
410 -- If the first list is not finite, the result is the first list.
412 (++) :: [a] -> [a] -> [a]
414 (++) (x:xs) ys = x : xs ++ ys
417 "++" [~1] forall xs ys. xs ++ ys = augment (\c n -> foldr c n xs) ys
423 %*********************************************************
425 \subsection{Type @Bool@}
427 %*********************************************************
430 -- |'otherwise' is defined as the value 'True'. It helps to make
431 -- guards more readable. eg.
433 -- > f x | x < 0 = ...
434 -- > | otherwise = ...
439 %*********************************************************
441 \subsection{Type @Char@ and @String@}
443 %*********************************************************
446 -- | A 'String' is a list of characters. String constants in Haskell are values
452 "x# `eqChar#` x#" forall x#. x# `eqChar#` x# = True
453 "x# `neChar#` x#" forall x#. x# `neChar#` x# = False
454 "x# `gtChar#` x#" forall x#. x# `gtChar#` x# = False
455 "x# `geChar#` x#" forall x#. x# `geChar#` x# = True
456 "x# `leChar#` x#" forall x#. x# `leChar#` x# = True
457 "x# `ltChar#` x#" forall x#. x# `ltChar#` x# = False
460 -- | The 'Prelude.toEnum' method restricted to the type 'Data.Char.Char'.
463 | int2Word# i# `leWord#` int2Word# 0x10FFFF# = C# (chr# i#)
465 = error ("Prelude.chr: bad argument: " ++ showSignedInt (I# 9#) i "")
467 unsafeChr :: Int -> Char
468 unsafeChr (I# i#) = C# (chr# i#)
470 -- | The 'Prelude.fromEnum' method restricted to the type 'Data.Char.Char'.
472 ord (C# c#) = I# (ord# c#)
475 String equality is used when desugaring pattern-matches against strings.
478 eqString :: String -> String -> Bool
479 eqString [] [] = True
480 eqString (c1:cs1) (c2:cs2) = c1 == c2 && cs1 `eqString` cs2
483 {-# RULES "eqString" (==) = eqString #-}
484 -- eqString also has a BuiltInRule in PrelRules.lhs:
485 -- eqString (unpackCString# (Lit s1)) (unpackCString# (Lit s2) = s1==s2
489 %*********************************************************
491 \subsection{Type @Int@}
493 %*********************************************************
496 zeroInt, oneInt, twoInt, maxInt, minInt :: Int
501 {- Seems clumsy. Should perhaps put minInt and MaxInt directly into MachDeps.h -}
502 #if WORD_SIZE_IN_BITS == 31
503 minInt = I# (-0x40000000#)
504 maxInt = I# 0x3FFFFFFF#
505 #elif WORD_SIZE_IN_BITS == 32
506 minInt = I# (-0x80000000#)
507 maxInt = I# 0x7FFFFFFF#
509 minInt = I# (-0x8000000000000000#)
510 maxInt = I# 0x7FFFFFFFFFFFFFFF#
513 instance Eq Int where
517 instance Ord Int where
524 compareInt :: Int -> Int -> Ordering
525 (I# x#) `compareInt` (I# y#) = compareInt# x# y#
527 compareInt# :: Int# -> Int# -> Ordering
535 %*********************************************************
537 \subsection{The function type}
539 %*********************************************************
542 -- | Identity function.
546 -- | The call '(lazy e)' means the same as 'e', but 'lazy' has a
547 -- magical strictness property: it is lazy in its first argument,
548 -- even though its semantics is strict.
551 -- Implementation note: its strictness and unfolding are over-ridden
552 -- by the definition in MkId.lhs; in both cases to nothing at all.
553 -- That way, 'lazy' does not get inlined, and the strictness analyser
554 -- sees it as lazy. Then the worker/wrapper phase inlines it.
557 -- Assertion function. This simply ignores its boolean argument.
558 -- The compiler may rewrite it to @('assertError' line)@.
560 -- | If the first argument evaluates to 'True', then the result is the
561 -- second argument. Otherwise an 'AssertionFailed' exception is raised,
562 -- containing a 'String' with the source file and line number of the
565 -- Assertions can normally be turned on or off with a compiler flag
566 -- (for GHC, assertions are normally on unless optimisation is turned on
567 -- with @-O@ or the @-fignore-asserts@
568 -- option is given). When assertions are turned off, the first
569 -- argument to 'assert' is ignored, and the second argument is
570 -- returned as the result.
572 -- SLPJ: in 5.04 etc 'assert' is in GHC.Prim,
573 -- but from Template Haskell onwards it's simply
574 -- defined here in Base.lhs
575 assert :: Bool -> a -> a
581 breakpointCond :: Bool -> a -> a
582 breakpointCond _ r = r
584 data Opaque = forall a. O a
586 -- | Constant function.
590 -- | Function composition.
592 -- Make sure it has TWO args only on the left, so that it inlines
593 -- when applied to two functions, even if there is no final argument
594 (.) :: (b -> c) -> (a -> b) -> a -> c
595 (.) f g = \x -> f (g x)
597 -- | @'flip' f@ takes its (first) two arguments in the reverse order of @f@.
598 flip :: (a -> b -> c) -> b -> a -> c
601 -- | Application operator. This operator is redundant, since ordinary
602 -- application @(f x)@ means the same as @(f '$' x)@. However, '$' has
603 -- low, right-associative binding precedence, so it sometimes allows
604 -- parentheses to be omitted; for example:
606 -- > f $ g $ h x = f (g (h x))
608 -- It is also useful in higher-order situations, such as @'map' ('$' 0) xs@,
609 -- or @'Data.List.zipWith' ('$') fs xs@.
611 ($) :: (a -> b) -> a -> b
614 -- | @'until' p f@ yields the result of applying @f@ until @p@ holds.
615 until :: (a -> Bool) -> (a -> a) -> a -> a
616 until p f x | p x = x
617 | otherwise = until p f (f x)
619 -- | 'asTypeOf' is a type-restricted version of 'const'. It is usually
620 -- used as an infix operator, and its typing forces its first argument
621 -- (which is usually overloaded) to have the same type as the second.
622 asTypeOf :: a -> a -> a
626 %*********************************************************
628 \subsection{@Functor@ and @Monad@ instances for @IO@}
630 %*********************************************************
633 instance Functor IO where
634 fmap f x = x >>= (return . f)
636 instance Monad IO where
637 {-# INLINE return #-}
640 m >> k = m >>= \ _ -> k
643 fail s = GHC.IO.failIO s
645 returnIO :: a -> IO a
646 returnIO x = IO $ \ s -> (# s, x #)
648 bindIO :: IO a -> (a -> IO b) -> IO b
649 bindIO (IO m) k = IO $ \ s -> case m s of (# new_s, a #) -> unIO (k a) new_s
651 thenIO :: IO a -> IO b -> IO b
652 thenIO (IO m) k = IO $ \ s -> case m s of (# new_s, _ #) -> unIO k new_s
654 unIO :: IO a -> (State# RealWorld -> (# State# RealWorld, a #))
658 %*********************************************************
660 \subsection{@getTag@}
662 %*********************************************************
664 Returns the 'tag' of a constructor application; this function is used
665 by the deriving code for Eq, Ord and Enum.
667 The primitive dataToTag# requires an evaluated constructor application
668 as its argument, so we provide getTag as a wrapper that performs the
669 evaluation before calling dataToTag#. We could have dataToTag#
670 evaluate its argument, but we prefer to do it this way because (a)
671 dataToTag# can be an inline primop if it doesn't need to do any
672 evaluation, and (b) we want to expose the evaluation to the
673 simplifier, because it might be possible to eliminate the evaluation
674 in the case when the argument is already known to be evaluated.
677 {-# INLINE getTag #-}
679 getTag x = x `seq` dataToTag# x
682 %*********************************************************
684 \subsection{Numeric primops}
686 %*********************************************************
689 divInt# :: Int# -> Int# -> Int#
691 -- Be careful NOT to overflow if we do any additional arithmetic
692 -- on the arguments... the following previous version of this
693 -- code has problems with overflow:
694 -- | (x# ># 0#) && (y# <# 0#) = ((x# -# y#) -# 1#) `quotInt#` y#
695 -- | (x# <# 0#) && (y# ># 0#) = ((x# -# y#) +# 1#) `quotInt#` y#
696 | (x# ># 0#) && (y# <# 0#) = ((x# -# 1#) `quotInt#` y#) -# 1#
697 | (x# <# 0#) && (y# ># 0#) = ((x# +# 1#) `quotInt#` y#) -# 1#
698 | otherwise = x# `quotInt#` y#
700 modInt# :: Int# -> Int# -> Int#
702 | (x# ># 0#) && (y# <# 0#) ||
703 (x# <# 0#) && (y# ># 0#) = if r# /=# 0# then r# +# y# else 0#
706 !r# = x# `remInt#` y#
709 Definitions of the boxed PrimOps; these will be
710 used in the case of partial applications, etc.
719 {-# INLINE plusInt #-}
720 {-# INLINE minusInt #-}
721 {-# INLINE timesInt #-}
722 {-# INLINE quotInt #-}
723 {-# INLINE remInt #-}
724 {-# INLINE negateInt #-}
726 plusInt, minusInt, timesInt, quotInt, remInt, divInt, modInt :: Int -> Int -> Int
727 (I# x) `plusInt` (I# y) = I# (x +# y)
728 (I# x) `minusInt` (I# y) = I# (x -# y)
729 (I# x) `timesInt` (I# y) = I# (x *# y)
730 (I# x) `quotInt` (I# y) = I# (x `quotInt#` y)
731 (I# x) `remInt` (I# y) = I# (x `remInt#` y)
732 (I# x) `divInt` (I# y) = I# (x `divInt#` y)
733 (I# x) `modInt` (I# y) = I# (x `modInt#` y)
736 "x# +# 0#" forall x#. x# +# 0# = x#
737 "0# +# x#" forall x#. 0# +# x# = x#
738 "x# -# 0#" forall x#. x# -# 0# = x#
739 "x# -# x#" forall x#. x# -# x# = 0#
740 "x# *# 0#" forall x#. x# *# 0# = 0#
741 "0# *# x#" forall x#. 0# *# x# = 0#
742 "x# *# 1#" forall x#. x# *# 1# = x#
743 "1# *# x#" forall x#. 1# *# x# = x#
746 negateInt :: Int -> Int
747 negateInt (I# x) = I# (negateInt# x)
749 gtInt, geInt, eqInt, neInt, ltInt, leInt :: Int -> Int -> Bool
750 (I# x) `gtInt` (I# y) = x ># y
751 (I# x) `geInt` (I# y) = x >=# y
752 (I# x) `eqInt` (I# y) = x ==# y
753 (I# x) `neInt` (I# y) = x /=# y
754 (I# x) `ltInt` (I# y) = x <# y
755 (I# x) `leInt` (I# y) = x <=# y
758 "x# ># x#" forall x#. x# ># x# = False
759 "x# >=# x#" forall x#. x# >=# x# = True
760 "x# ==# x#" forall x#. x# ==# x# = True
761 "x# /=# x#" forall x#. x# /=# x# = False
762 "x# <# x#" forall x#. x# <# x# = False
763 "x# <=# x#" forall x#. x# <=# x# = True
767 "plusFloat x 0.0" forall x#. plusFloat# x# 0.0# = x#
768 "plusFloat 0.0 x" forall x#. plusFloat# 0.0# x# = x#
769 "minusFloat x 0.0" forall x#. minusFloat# x# 0.0# = x#
770 "minusFloat x x" forall x#. minusFloat# x# x# = 0.0#
771 "timesFloat x 0.0" forall x#. timesFloat# x# 0.0# = 0.0#
772 "timesFloat0.0 x" forall x#. timesFloat# 0.0# x# = 0.0#
773 "timesFloat x 1.0" forall x#. timesFloat# x# 1.0# = x#
774 "timesFloat 1.0 x" forall x#. timesFloat# 1.0# x# = x#
775 "divideFloat x 1.0" forall x#. divideFloat# x# 1.0# = x#
779 "plusDouble x 0.0" forall x#. (+##) x# 0.0## = x#
780 "plusDouble 0.0 x" forall x#. (+##) 0.0## x# = x#
781 "minusDouble x 0.0" forall x#. (-##) x# 0.0## = x#
782 "timesDouble x 1.0" forall x#. (*##) x# 1.0## = x#
783 "timesDouble 1.0 x" forall x#. (*##) 1.0## x# = x#
784 "divideDouble x 1.0" forall x#. (/##) x# 1.0## = x#
788 We'd like to have more rules, but for example:
790 This gives wrong answer (0) for NaN - NaN (should be NaN):
791 "minusDouble x x" forall x#. (-##) x# x# = 0.0##
793 This gives wrong answer (0) for 0 * NaN (should be NaN):
794 "timesDouble 0.0 x" forall x#. (*##) 0.0## x# = 0.0##
796 This gives wrong answer (0) for NaN * 0 (should be NaN):
797 "timesDouble x 0.0" forall x#. (*##) x# 0.0## = 0.0##
799 These are tested by num014.
802 -- Wrappers for the shift operations. The uncheckedShift# family are
803 -- undefined when the amount being shifted by is greater than the size
804 -- in bits of Int#, so these wrappers perform a check and return
805 -- either zero or -1 appropriately.
807 -- Note that these wrappers still produce undefined results when the
808 -- second argument (the shift amount) is negative.
810 -- | Shift the argument left by the specified number of bits
811 -- (which must be non-negative).
812 shiftL# :: Word# -> Int# -> Word#
813 a `shiftL#` b | b >=# WORD_SIZE_IN_BITS# = int2Word# 0#
814 | otherwise = a `uncheckedShiftL#` b
816 -- | Shift the argument right by the specified number of bits
817 -- (which must be non-negative).
818 shiftRL# :: Word# -> Int# -> Word#
819 a `shiftRL#` b | b >=# WORD_SIZE_IN_BITS# = int2Word# 0#
820 | otherwise = a `uncheckedShiftRL#` b
822 -- | Shift the argument left by the specified number of bits
823 -- (which must be non-negative).
824 iShiftL# :: Int# -> Int# -> Int#
825 a `iShiftL#` b | b >=# WORD_SIZE_IN_BITS# = 0#
826 | otherwise = a `uncheckedIShiftL#` b
828 -- | Shift the argument right (signed) by the specified number of bits
829 -- (which must be non-negative).
830 iShiftRA# :: Int# -> Int# -> Int#
831 a `iShiftRA#` b | b >=# WORD_SIZE_IN_BITS# = if a <# 0# then (-1#) else 0#
832 | otherwise = a `uncheckedIShiftRA#` b
834 -- | Shift the argument right (unsigned) by the specified number of bits
835 -- (which must be non-negative).
836 iShiftRL# :: Int# -> Int# -> Int#
837 a `iShiftRL#` b | b >=# WORD_SIZE_IN_BITS# = 0#
838 | otherwise = a `uncheckedIShiftRL#` b
840 #if WORD_SIZE_IN_BITS == 32
842 "narrow32Int#" forall x#. narrow32Int# x# = x#
843 "narrow32Word#" forall x#. narrow32Word# x# = x#
848 "int2Word2Int" forall x#. int2Word# (word2Int# x#) = x#
849 "word2Int2Word" forall x#. word2Int# (int2Word# x#) = x#
854 %********************************************************
856 \subsection{Unpacking C strings}
858 %********************************************************
860 This code is needed for virtually all programs, since it's used for
861 unpacking the strings of error messages.
864 unpackCString# :: Addr# -> [Char]
865 {-# NOINLINE unpackCString# #-}
866 -- There's really no point in inlining this, ever, cos
867 -- the loop doesn't specialise in an interesting
868 -- But it's pretty small, so there's a danger that
869 -- it'll be inlined at every literal, which is a waste
874 | ch `eqChar#` '\0'# = []
875 | otherwise = C# ch : unpack (nh +# 1#)
877 !ch = indexCharOffAddr# addr nh
879 unpackAppendCString# :: Addr# -> [Char] -> [Char]
880 {-# NOINLINE unpackAppendCString# #-}
881 -- See the NOINLINE note on unpackCString#
882 unpackAppendCString# addr rest
886 | ch `eqChar#` '\0'# = rest
887 | otherwise = C# ch : unpack (nh +# 1#)
889 !ch = indexCharOffAddr# addr nh
891 unpackFoldrCString# :: Addr# -> (Char -> a -> a) -> a -> a
893 -- Usually the unpack-list rule turns unpackFoldrCString# into unpackCString#
895 -- It also has a BuiltInRule in PrelRules.lhs:
896 -- unpackFoldrCString# "foo" c (unpackFoldrCString# "baz" c n)
897 -- = unpackFoldrCString# "foobaz" c n
899 {-# NOINLINE unpackFoldrCString# #-}
900 -- At one stage I had NOINLINE [0] on the grounds that, unlike
901 -- unpackCString#, there *is* some point in inlining
902 -- unpackFoldrCString#, because we get better code for the
903 -- higher-order function call. BUT there may be a lot of
904 -- literal strings, and making a separate 'unpack' loop for
905 -- each is highly gratuitous. See nofib/real/anna/PrettyPrint.
907 unpackFoldrCString# addr f z
911 | ch `eqChar#` '\0'# = z
912 | otherwise = C# ch `f` unpack (nh +# 1#)
914 !ch = indexCharOffAddr# addr nh
916 unpackCStringUtf8# :: Addr# -> [Char]
917 unpackCStringUtf8# addr
921 | ch `eqChar#` '\0'# = []
922 | ch `leChar#` '\x7F'# = C# ch : unpack (nh +# 1#)
923 | ch `leChar#` '\xDF'# =
924 C# (chr# (((ord# ch -# 0xC0#) `uncheckedIShiftL#` 6#) +#
925 (ord# (indexCharOffAddr# addr (nh +# 1#)) -# 0x80#))) :
927 | ch `leChar#` '\xEF'# =
928 C# (chr# (((ord# ch -# 0xE0#) `uncheckedIShiftL#` 12#) +#
929 ((ord# (indexCharOffAddr# addr (nh +# 1#)) -# 0x80#) `uncheckedIShiftL#` 6#) +#
930 (ord# (indexCharOffAddr# addr (nh +# 2#)) -# 0x80#))) :
933 C# (chr# (((ord# ch -# 0xF0#) `uncheckedIShiftL#` 18#) +#
934 ((ord# (indexCharOffAddr# addr (nh +# 1#)) -# 0x80#) `uncheckedIShiftL#` 12#) +#
935 ((ord# (indexCharOffAddr# addr (nh +# 2#)) -# 0x80#) `uncheckedIShiftL#` 6#) +#
936 (ord# (indexCharOffAddr# addr (nh +# 3#)) -# 0x80#))) :
939 !ch = indexCharOffAddr# addr nh
941 unpackNBytes# :: Addr# -> Int# -> [Char]
942 unpackNBytes# _addr 0# = []
943 unpackNBytes# addr len# = unpack [] (len# -# 1#)
948 case indexCharOffAddr# addr i# of
949 ch -> unpack (C# ch : acc) (i# -# 1#)
952 "unpack" [~1] forall a . unpackCString# a = build (unpackFoldrCString# a)
953 "unpack-list" [1] forall a . unpackFoldrCString# a (:) [] = unpackCString# a
954 "unpack-append" forall a n . unpackFoldrCString# a (:) n = unpackAppendCString# a n
956 -- There's a built-in rule (in PrelRules.lhs) for
957 -- unpackFoldr "foo" c (unpackFoldr "baz" c n) = unpackFoldr "foobaz" c n
964 -- | A special argument for the 'Control.Monad.ST.ST' type constructor,
965 -- indexing a state embedded in the 'Prelude.IO' monad by
966 -- 'Control.Monad.ST.stToIO'.