Cryptol.Backend.Monad

Evaluation monad

data Eval a

The monad for Cryptol evaluation. A computation is either "ready", which means it represents only trivial computation, or is an "eval" action which must be computed to get the answer, or it is a "thunk", which represents a delayed, shared computation.

Constructors

Ready !a

Eval !(IO a)

Thunk !(TVar (ThunkState a))

Instances

Monad Eval

Defined in Cryptol.Backend.Monad

Methods

(>>=) :: Eval a -> (a -> Eval b) -> Eval b

(>>) :: Eval a -> Eval b -> Eval b

return :: a -> Eval a

Functor Eval

Defined in Cryptol.Backend.Monad

Methods

fmap :: (a -> b) -> Eval a -> Eval b

(<$) :: a -> Eval b -> Eval a

MonadFix Eval

Defined in Cryptol.Backend.Monad

Methods

mfix :: (a -> Eval a) -> Eval a

MonadFail Eval

Defined in Cryptol.Backend.Monad

Methods

fail :: String -> Eval a

Applicative Eval

Defined in Cryptol.Backend.Monad

Methods

pure :: a -> Eval a

(<*>) :: Eval (a -> b) -> Eval a -> Eval b

liftA2 :: (a -> b -> c) -> Eval a -> Eval b -> Eval c

(*>) :: Eval a -> Eval b -> Eval b

(<*) :: Eval a -> Eval b -> Eval a

MonadIO Eval

Defined in Cryptol.Backend.Monad

Methods

liftIO :: IO a -> Eval a

runEval :: Eval a -> IO a

Execute the given evaluation action.

data EvalOpts

Some options for evalutaion

Constructors

EvalOpts

Fields

• evalLogger :: Logger

  Where to print stuff (e.g., for trace)

• evalPPOpts :: PPOpts

  How to pretty print things.

data PPOpts

How to pretty print things when evaluating

Constructors

PPOpts

Fields

• useAscii :: Bool
• useBase :: Int
• useInfLength :: Int
• useFPBase :: Int
• useFPFormat :: PPFloatFormat

asciiMode :: PPOpts -> Integer -> Bool

data PPFloatFormat

Constructors

FloatFixed Int PPFloatExp

Use this many significant digis

FloatFrac Int

Show this many digits after floating point

FloatFree PPFloatExp

Use the correct number of digits

data PPFloatExp

Constructors

ForceExponent

Always show an exponent

AutoExponent

Only show exponent when needed

defaultPPOpts :: PPOpts

io :: IO a -> Eval a

Lift an IO computation into the Eval monad.

delayFill

Arguments

:: Eval a

Computation to delay

-> Eval a

Backup computation to run if a tight loop is detected

-> Eval (Eval a)

Delay the given evaluation computation, returning a thunk which will run the computation when forced. Run the retry computation instead if the resulting thunk is forced during its own evaluation.

ready :: a -> Eval a

A computation that returns an already-evaluated value.

blackhole

Arguments

:: String

A name to associate with this thunk.

-> Eval (Eval a, Eval a -> Eval ())

Produce a thunk value which can be filled with its associated computation after the fact. A preallocated thunk is returned, along with an operation to fill the thunk with the associated computation. This is used to implement recursive declaration groups.

evalSpark :: Eval a -> Eval (Eval a)

Begin executing the given operation in a separate thread, returning a thunk which will await the completion of the computation when forced.

maybeReady :: Eval a -> Eval (Maybe a)

Test if a value is "ready", which means that it requires no computation to return.

Error reporting

data Unsupported

Constructors

UnsupportedSymbolicOp String

Operation cannot be supported in the symbolic simulator

Instances

Show Unsupported

Defined in Cryptol.Backend.Monad

Methods

showsPrec :: Int -> Unsupported -> ShowS

show :: Unsupported -> String

showList :: [Unsupported] -> ShowS

Exception Unsupported

Defined in Cryptol.Backend.Monad

Methods

toException :: Unsupported -> SomeException

fromException :: SomeException -> Maybe Unsupported

displayException :: Unsupported -> String

PP Unsupported

Defined in Cryptol.Backend.Monad

Methods

ppPrec :: Int -> Unsupported -> Doc

data EvalError

Data type describing errors that can occur during evaluation.

Constructors

InvalidIndex (Maybe Integer)

Out-of-bounds index

TypeCannotBeDemoted Type

Non-numeric type passed to number function

DivideByZero

Division or modulus by 0

NegativeExponent

Exponentiation by negative integer

LogNegative

Logarithm of a negative integer

WordTooWide Integer

Bitvector too large

UserError String

Call to the Cryptol error primitive

LoopError String

Detectable nontermination

NoPrim Name

Primitive with no implementation

BadRoundingMode Integer

Invalid rounding mode

BadValue String

Value outside the domain of a partial function.

Instances

Show EvalError

Defined in Cryptol.Backend.Monad

Methods

showsPrec :: Int -> EvalError -> ShowS

show :: EvalError -> String

showList :: [EvalError] -> ShowS

Exception EvalError

Defined in Cryptol.Backend.Monad

Methods

toException :: EvalError -> SomeException

fromException :: SomeException -> Maybe EvalError

displayException :: EvalError -> String

PP EvalError

Defined in Cryptol.Backend.Monad

Methods

ppPrec :: Int -> EvalError -> Doc

evalPanic :: HasCallStack => String -> [String] -> a

Panic from an Eval context.

wordTooWide :: Integer -> a

For when we know that a word is too wide and will exceed gmp's limits (though words approaching this size will probably cause the system to crash anyway due to lack of memory).

typeCannotBeDemoted :: Type -> a

For things like `(inf) or `(0-1).