Skip to content

Latest commit

 

History

History
314 lines (240 loc) · 9.94 KB

README.md

File metadata and controls

314 lines (240 loc) · 9.94 KB
Raw

Disclaimer: The provided software is a developing prototype in its early days, not a released product. It is provided only to be evaluated together with Venom Hackathon submission. Lots of details described in this README are planned features, not yet implemented

Light Programming Language

   __    _       __    __
   / /   (_)___ _/ /_  / /_
  / /   / / __ `/ __ \/ __/
 / /___/ / /_/ / / / / /_
/_____/_/\__, /_/ /_/\__/
        /____/

Light is a next-generation programming language targeting TON-inspired blockchains, and specifically EverScale and Venom. Light is a statically-typed functional reactive actor-based programming language with lots of features coming for the first time ever in a modern blockchain programming landscape.

Light is a part of the bigger Lighthouse programming system aiming to significantly boost developers productivity and program safety, reducing time-to-market and lowering project delivery risks. For details, have a look at our Lighthouse Whitepaper.

Language Features

We highlight the following features of the Light language:

  • Algebraic Data Types with Type Parameters

    "Product" types for records, and "sum"-types for variants, recursive type definitions, etc. This type system provides means for specifying rich message and storage data schemes.

    (* State type defines the content and shape of the actor state *)
type State = {
  pubkey: PubKey ;
  deposits: map<ActorId, uint> ;
  requests: list<ActorId>
}

(* Sum (or Variant) type define several possible constructors
   for values; you may use only one of them *)
type Token =
| FungibleToken of n:uint
| NonFungibleToken of id:ActorId

(* Recursive data types *)
type List<'T> =
| Nil
| Cons of (h : T) * (t : List<T>)

(* Optional type *)
type Option<'T> =
| None
| Some of (n:T)

  • Functions as First-Class Objects

    The really distinguishing feature of the Light language is its ability to send, receive, store, extend and shrink arbitrary functions. It is for the first time ever you can not only transmit data, but also attach custom business logic to it. Computation can now travel between systems of actors gradually converging into a result.

    (* Attach functions to messages as if it is an ordinary data. *)
(* Later, the receiver will be able to execute the function   *)
type Message =
  Message of (n:uint) * (func: uint -> uint)

(* Recursive factorial function *)
let rec fac n =
    if (n > 1) then n * fac (n - 1)
    else 1
;;

let sum a b = a + b ;;        (* sum of two numbers, just for example *)
let sum5 = sum 5 ;;           (* partial function application *)
let sum105 = sum5 100 ;;      (* this equals 5 + 100 *)

(* Tail-recursive sum of numbers list *)
(* example: sum_list2 [1;2;3] 0  =  6   *)
let rec sum_list2 lst acc =
   match lst with
   | [] -> acc
   | h :: t -> sum_list t (h + acc)

  • Message-oriented control flow

    The language provides several operators that allow structuring the control flow around the message processing.

    • operator receive
    • operator send (!)

    Operator receive gives opportunity to express logic in a form of dialogues between smart-contracts.

    receive
| Deposit (NonFungibleTokens n) ->
  if n < 100 then
    let sender = ctx.msg.src in
    sender ! DepositFeedback (SendExtraTokens (100 - n)) ;
    receive
    | Deposit (NonFungibleTokens p) when ctx.msg.src = sender  ->
       assert (n + p > 100) ;
       depositTokens ctx.msg.src (n + p)
    after 100 ->
       failwith #"extra deposit request failed"
after 100 ->
  failwith #"amount is too small"
end

    The receive operator "suspends" actor execution until a new message arrives; after that, the actor does the pattern-matching on the message content. If the corresponding pattern is found, the body of the corresponding pattern handler is executed. Unmatched messages may be skipped or bounced back. The "after" value specifies the message receiving deadline, i.e. for how long we wait for the message before giving up and continuing the execution. The operator "receive" can be nested.

  • Optional Mutability

    Functional programming encourages programming with immutable variables, but when you need mutability, you still have it, with familiar cycles, exceptions etc.

    try
  let mutable cond = false in
  let mutable n = 0 in
  while not cond do
    n <- n + 1 ;
    if (n > 100) then
       cond <- true ;
    if (n > 1000) then
       raise (Overflow n) ;
  done
with
| Overflow k ->
    // specific exception handler goes here
| e ->
    // general exception handler goes here

  • Actors as First-class Objects

    The language provides means to spawn ('create') other actors on the fly, with convenient programming constructs. The actor logic is passed either as module definition or as a function.

    type ActorMessage =
| Say of (s:string)
| Quit

(* This is a body of another actor *)
let rec actorFun () =
  receive
  | Say of s ->
     print s ;
     (* recursive call to itself after message received *)
     actorFun ()
  | Quit -> ()

(* create new actor with empty state *)
let actId = spawn actFun () in
actId  ! Say "hi!" ;         (* Send some messages *)
actId  ! Say "hello!" ;
actId  ! Quit

    Spawning actors "in-place" is also possible:

    spawn ( fun () -> ctx.sender ! ("hi!") ) ;

  • Static Typing with Automatic Inference

    If the program compiles, it is considered safe (runtime exception safety) most of the time. Strong static type system safe programmers from subtle bugs. However, specifying types may be daunting task. This is why Light provide automatic type inference.

    Types are automatically inferred from the context. The type safety is guaranteed by the compiler.

  let print_string s =
     print s

  (* this is also correct *)
  let print_string (s:string) =
     print s

  • Delayed (Lazy) Execution

    Some pieces of logic should be executed only when and if they are needed, not when they are defined. This lets you define things without worrying too much about sub-optimal gas usage. This, in turn, removes the temptation of doing premature optimization that is, as we all know, the root of all evil.

    Lazy computation construct allows exactly this, in a type-safe way.

    (* We do not know how many sequence numbers we will
   need in the future. This is why we generate infinite
   stream of them, taking one at a time when we need them. *)
let nextNumber () : Lazy<uint> =
  let mutable n:uint = 0 in
  while true do
    n <- n + 1 ;
    yield n
  done
in
  let x = force (nextNumber ()) in   (* x = 1 *)
  let y = force (nextNumber ()) in   (* y = 2 *)
  x + y   (* =3 *)

Prototype restrictions

The vision stated in the Overview section implemented only partially at the moment. Currently, we implemented:

  • Compiler for Light Core - the core language that lets us build all the described constructs atop of it.
  • Deploy scripts that let you prepare .BOC files to deploy actors in the blockchain
  • Message scripts that let you prepare .BOC with messages
  • Prepared samples in samples folder.

Manual

Here we provide step-by-step guide how to make the compiler prototype work for you.

Light compiler rely on STCONT/LDCONT instructions. Both instructions present in Venom/EverScale VM, however, its availability for smart-contracts is regulated by the network capability CapStContNewFormat.

Currently, Venom Dev-net has this cap turned off, so there are two options left:

  1. Run actors locally using tvm_linker test (not that fun, but still a viable option)
  2. Run actors in FLD network: the FLD network administrator kindly enabled this capability specifically for Light programmers.

Installation

1. Build and install the following:

  • tvm_linker from here (NOTE: it is our custom tvm_linker, not the one supplied by EverX!)

  • fift from here

  • tonos-cli from here

After installation, ensure that all commands are visible inside your $PATH.

2. Install Microsoft .NET framework.

See here for instruction on how to do that for your Linux distro.

Ensure that dotnet fsi command is working.

3. Build the Light compiler.

$ git clone https://github.com/unboxedtype/light
$ cd light
$ make build

4. Put the directory <light>/scripts/ into the PATH

$ export PATH=$PATH:$(pwd)/scripts/

Check that the command genActorMessage.fsx and serializeExpression.fsx are visible.

5. Make the LHCompiler binary visible. For that, do one of the following:

  • Put the directory <light>/src/LHCompiler/bin/net6.0/ into the PATH

    $ export PATH=$PATH:$(pwd)/src/LHCompiler/bin/net6.0/

  • OR Make a symbolic link to LHCompiler binary:

    $ sudo ln -s $(pwd)/src/LHCompiler/bin/Debug/net6.0/LHCompiler /usr/bin/LHCompiler

Ensure that the command LHCompiler works afterwards.

6. Go to <light>/samples/Sample<N> directory. There you will find test.sh script. Run it and do what it asks for. Inside the scripts, you can find all the necessary commands to deploy and interact with Light actors!

Community

Contact @unboxedType on Telegram for questions and suggestions.