Skip to main content

Algebraic Effects

Note that this feature is still in an experimental stage.

Miking's standard library includes support for working with effectful computations.

The relevant file is stdlib/effect.mc, inspired by algebraic effects / free(er) monads. To learn more, see e.g., 1,2,3.

Overview

The main part of the library is the Effect fragment, which defines a type Eff a representing generic effectful computations. Users may extend Eff with their own effectful operations along with handlers for those operations.

Effects interact well with language fragment composition. To use a given effect, simply include the corresponding fragment and start using the effectful operations. A sem clause in one language fragment may use a given effect (e.g., a read-only context) without interfering with the definitions of other clauses (e.g., without having to add an extra parameter even in clauses where the context is unused). The only requirement is that the sem returns an Eff-type.

In addition to stdlib/effect.mc, there is also stdlib/mexpr/ast-effect.mc, which contains variations of an smapEff-function which can be used to map effectful operations over MExpr AST nodes.

A small example

For example, consider the following fragment:

lang TestLang = Reader + NonDet
sem getInt : Ctx -> Int

sem effProg : () -> Eff Int
sem effProg = | () ->
bind (choose [0,1]) (lam i.
bind (choose [2,3]) (lam j.
bind (ask getInt) (lam k.
return (addi (addi i j) k))))

end

The fragment TestLang imports two effects: read-only context (Reader) and non-determinism (NonDet). effProg defines an effectful program which uses both effects. The program first chooses two numbers i \in [0,1] and j \in [2,3], then reads a number k from the context, and then returns the sum of the three. Note that the context type is kept abstract, we only declare a sem getInt : Ctx -> Int for extracting an integer from it.

To run the program, we provide a concrete implementation of getInt and a value for the context:

lang TestLangImpl = TestLang
syn Ctx = | ICtx Int
sem getInt = | ICtx i -> i
end

mexpr

use TestLangImpl in

utest runEff (handleND (handleReader (ICtx 7) (effProg ())))
with [9,10,10,11] in ()

Here, the Reader effect is first handled using handleReader with a context ICtx 7, and then the nondeterminism effect is resolved using handleND. When all effects are handled, runEff can be used to extract the final value, in this case [9,10,10,11].

A larger example

Now, we show a larger example demonstrating an interpreter for the lambda calculus plus integers and addition written modularly using MLang's language fragments. The interpreter is structured into two fragments: the Arith fragment which defines the semantics for integers and addition, and the Lam fragment which defines the semantics for variables, functions and applications. Note how Arith does not deal with variables at all, and yet it can be seamlessly composed with Lam, which (implicitly) needs to pass around a variable context using the Reader effect.

include "common.mc"
include "effect.mc"
include "map.mc"

lang Eval
syn Expr =
syn Value =

sem eval : Expr -> Eff Value
end

lang Arith = Eval + Failure
syn Expr =
| Lit Int
| Add (Expr, Expr)

syn Value =
| VInt Int

syn Failure =
| InvalidAddition ()

sem evalAdd : (Value, Value) -> Eff Value
sem evalAdd =
| (VInt i, VInt j) -> return (VInt (addi i j))
| _ -> fail (InvalidAddition ())

sem eval =
| Lit i -> return (VInt i)
| Add (e1, e2) ->
bind (eval e1) (lam v1.
bind (eval e2) (lam v2.
evalAdd (v1, v2)))
end


lang Lam = Eval + Reader + Failure
syn Expr =
| Lam (String, Expr)
| Var String
| App (Expr, Expr)

syn Value =
| VClos (Map String Value, String, Expr)

syn Failure =
| MissingVariable String
| InvalidApplication ()

sem getVarCtx : Ctx -> Map String Value
sem putVarCtx : Map String Value -> Ctx -> Ctx

sem eval =
| Var s ->
bind (ask getVarCtx) (lam ctx.
match mapLookup s ctx with Some v then return v
else fail (MissingVariable s))
| Lam (s, e) ->
bind (ask getVarCtx) (lam ctx.
return (VClos (ctx, s, e)))
| App (e1, e2) ->
bind (eval e1) (lam v1.
bind (eval e2) (lam v2.
match v1 with VClos (ctx, s, e) then
local (putVarCtx (mapInsert s v2 ctx)) (eval e)
else fail (InvalidApplication ())))
end


lang EvalClient = Arith + Lam
syn Ctx = | Vars (Map String Value)

sem getVarCtx =
| Vars ctx -> ctx
sem putVarCtx ctx =
| _ -> Vars ctx
end

mexpr


use EvalClient in

let formatFailure : Failure -> String = lam f.
switch f
case InvalidAddition () then "Attempted to add non-integer values!"
case InvalidApplication () then "Attempted to apply non-function value!"
case MissingVariable s then join ["Unknown variable ", s, " encountered!"]
end
in

let formatValue : Value -> String = lam v.
switch v
case VInt i then int2string i
case VClos _ then "<function>"
end
in

let e : Expr =
App (Lam ("x", Add (Var "x", Lit 5)), Lit 7)
in

let result : Either String Value =
runEff
(handleFail formatFailure
(handleReader (Vars (mapEmpty cmpString))
(eval e)))
in

let output =
switch result
case Right v then formatValue v
case Left s then join ["ERROR: ", s]
end
in

printLn output

Limitations

There are three main limitations of the implementation at current. The first is that effectful computations have to be written in explicit monadic style as seen above. The second is that there is no static check that all effects have been handled. Trying to extract a value from an effectful computation without handling all effects will give a runtime error. The third limitation is that the runtime performance is not very good. There are techniques for improving the efficiency (e.g., 3), but these are not implemented currently.