Copyright | (c) Roman Leshchinskiy 2008-2010 |
---|---|
License | BSD-style |
Maintainer | Roman Leshchinskiy <rl@cse.unsw.edu.au> |
Stability | experimental |
Portability | non-portable |
Safe Haskell | None |
Language | Haskell2010 |
Mutable primitive vectors.
Synopsis
- data MVector s a = MVector !Int !Int !(MutableByteArray s)
- type IOVector = MVector RealWorld
- type STVector s = MVector s
- class Prim a
- length :: Prim a => MVector s a -> Int
- null :: Prim a => MVector s a -> Bool
- slice :: Prim a => Int -> Int -> MVector s a -> MVector s a
- init :: Prim a => MVector s a -> MVector s a
- tail :: Prim a => MVector s a -> MVector s a
- take :: Prim a => Int -> MVector s a -> MVector s a
- drop :: Prim a => Int -> MVector s a -> MVector s a
- splitAt :: Prim a => Int -> MVector s a -> (MVector s a, MVector s a)
- unsafeSlice :: Prim a => Int -> Int -> MVector s a -> MVector s a
- unsafeInit :: Prim a => MVector s a -> MVector s a
- unsafeTail :: Prim a => MVector s a -> MVector s a
- unsafeTake :: Prim a => Int -> MVector s a -> MVector s a
- unsafeDrop :: Prim a => Int -> MVector s a -> MVector s a
- overlaps :: Prim a => MVector s a -> MVector s a -> Bool
- new :: (PrimMonad m, Prim a) => Int -> m (MVector (PrimState m) a)
- unsafeNew :: (PrimMonad m, Prim a) => Int -> m (MVector (PrimState m) a)
- replicate :: (PrimMonad m, Prim a) => Int -> a -> m (MVector (PrimState m) a)
- replicateM :: (PrimMonad m, Prim a) => Int -> m a -> m (MVector (PrimState m) a)
- clone :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> m (MVector (PrimState m) a)
- grow :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m (MVector (PrimState m) a)
- unsafeGrow :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m (MVector (PrimState m) a)
- clear :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> m ()
- read :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m a
- write :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> a -> m ()
- modify :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> (a -> a) -> Int -> m ()
- swap :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> Int -> m ()
- unsafeRead :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m a
- unsafeWrite :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> a -> m ()
- unsafeModify :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> (a -> a) -> Int -> m ()
- unsafeSwap :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> Int -> m ()
- nextPermutation :: (PrimMonad m, Ord e, Prim e) => MVector (PrimState m) e -> m Bool
- set :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> a -> m ()
- copy :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> MVector (PrimState m) a -> m ()
- move :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> MVector (PrimState m) a -> m ()
- unsafeCopy :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> MVector (PrimState m) a -> m ()
- unsafeMove :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> MVector (PrimState m) a -> m ()
Mutable vectors of primitive types
Mutable vectors of primitive types.
MVector !Int !Int !(MutableByteArray s) | offset, length, underlying mutable byte array |
Instances
Class of types supporting primitive array operations. This includes
interfacing with GC-managed memory (functions suffixed with ByteArray#
)
and interfacing with unmanaged memory (functions suffixed with Addr#
).
Endianness is platform-dependent.
sizeOf#, alignment#, indexByteArray#, readByteArray#, writeByteArray#, setByteArray#, indexOffAddr#, readOffAddr#, writeOffAddr#, setOffAddr#
Instances
Accessors
Length information
Extracting subvectors
Yield a part of the mutable vector without copying it. The vector must
contain at least i+n
elements.
Yield a part of the mutable vector without copying it. No bounds checks are performed.
Overlapping
Construction
Initialisation
new :: (PrimMonad m, Prim a) => Int -> m (MVector (PrimState m) a) Source #
Create a mutable vector of the given length.
unsafeNew :: (PrimMonad m, Prim a) => Int -> m (MVector (PrimState m) a) Source #
Create a mutable vector of the given length. The vector content is uninitialized, which means it is filled with whatever underlying memory buffer happens to contain.
Since: 0.5
replicate :: (PrimMonad m, Prim a) => Int -> a -> m (MVector (PrimState m) a) Source #
Create a mutable vector of the given length (0 if the length is negative) and fill it with an initial value.
replicateM :: (PrimMonad m, Prim a) => Int -> m a -> m (MVector (PrimState m) a) Source #
Create a mutable vector of the given length (0 if the length is negative) and fill it with values produced by repeatedly executing the monadic action.
clone :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> m (MVector (PrimState m) a) Source #
Create a copy of a mutable vector.
Growing
grow :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m (MVector (PrimState m) a) Source #
Grow a primitive vector by the given number of elements. The number must be
non-negative. Same semantics as in grow
for generic vector.
Examples
>>>
import qualified Data.Vector.Primitive as VP
>>>
import qualified Data.Vector.Primitive.Mutable as MVP
>>>
mv <- VP.thaw $ VP.fromList ([10, 20, 30] :: [Int])
>>>
mv' <- MVP.grow mv 2
Extra memory at the end of the newly allocated vector is initialized to 0
bytes, which for Prim
instance will usually correspond to some default
value for a particular type, eg. 0
for Int
, NUL
for Char
,
etc. However, if unsafeGrow
was used instead this would not have been
guaranteed and some garbage would be there instead:
>>>
VP.unsafeFreeze mv'
[10,20,30,0,0]
Having the extra space we can write new values in there:
>>>
MVP.write mv' 3 999
>>>
VP.unsafeFreeze mv'
[10,20,30,999,0]
It is important to note that the source mutable vector is not affected when the newly allocated one is mutated.
>>>
MVP.write mv' 2 888
>>>
VP.unsafeFreeze mv'
[10,20,888,999,0]>>>
VP.unsafeFreeze mv
[10,20,30]
Since: 0.5
unsafeGrow :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m (MVector (PrimState m) a) Source #
Grow a vector by the given number of elements. The number must be non-negative but
this is not checked. Same semantics as in unsafeGrow
for generic vector.
Since: 0.5
Restricting memory usage
clear :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> m () Source #
Reset all elements of the vector to some undefined value, clearing all references to external objects. This is usually a noop for unboxed vectors.
Accessing individual elements
read :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m a Source #
Yield the element at the given position.
write :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> a -> m () Source #
Replace the element at the given position.
modify :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> (a -> a) -> Int -> m () Source #
Modify the element at the given position.
swap :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> Int -> m () Source #
Swap the elements at the given positions.
unsafeRead :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> m a Source #
Yield the element at the given position. No bounds checks are performed.
unsafeWrite :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> a -> m () Source #
Replace the element at the given position. No bounds checks are performed.
unsafeModify :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> (a -> a) -> Int -> m () Source #
Modify the element at the given position. No bounds checks are performed.
unsafeSwap :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> Int -> Int -> m () Source #
Swap the elements at the given positions. No bounds checks are performed.
Modifying vectors
nextPermutation :: (PrimMonad m, Ord e, Prim e) => MVector (PrimState m) e -> m Bool Source #
Compute the next (lexicographically) permutation of given vector in-place. Returns False when input is the last permutation
Filling and copying
set :: (PrimMonad m, Prim a) => MVector (PrimState m) a -> a -> m () Source #
Set all elements of the vector to the given value.
Copy a vector. The two vectors must have the same length and may not overlap.
Move the contents of a vector. The two vectors must have the same length.
If the vectors do not overlap, then this is equivalent to copy
.
Otherwise, the copying is performed as if the source vector were
copied to a temporary vector and then the temporary vector was copied
to the target vector.
Copy a vector. The two vectors must have the same length and may not overlap. This is not checked.
Move the contents of a vector. The two vectors must have the same length, but this is not checked.
If the vectors do not overlap, then this is equivalent to unsafeCopy
.
Otherwise, the copying is performed as if the source vector were
copied to a temporary vector and then the temporary vector was copied
to the target vector.