Oblivia: Object-Building Language Intended for Very Idiosyncratic Articulations

Oblivia (OBL) is an esolang that aims to do with objects what Lisp does with lists. Oblivia follows these ideas:

• Terse syntax:
  • A small set of primitive operations are given the most terse syntax.
  • Casting values is as easy as A(B) with type A and value B.
  • Defining variables is simply A:B(C) with variable name A, type B, and value C.
• Scopes = Objects: In Oblivia, scopes have the power of objects.
  • Pass a scope to a function.
  • Name a variable or a member and it automatically becomes a key of the scope.
  • A scope with no explicit return simply returns itself as an object (if it has keys) or the last expression (no keys).
• Objects = Functions: In Oblivia, any object can define the (), [], {} operators.
• Variables of any name: Variable names are not restricted by those already defined in the outer scope. To access variables of an outer scope, Oblivia introduces the ^ "up" function
• Same syntax everywhere:
  • : is the define operator
  • A(B), A[B], A{C} is a function call.
  • Patterns can be stored in variables: Some$(val)
• Classes are objects too: Classes can have a static companion (or not) that implements interfaces and behaves just like regular singleton objects.
• Term and Expression scoping: The right hand side of an operator has either term scope or expression scope. A term is a single item and an expression is a sequence of terms connected by operators. A term is one of the following.
  • A name
  • A number
  • A string
  • A tuple
  • An array
  • An object
  • A monadic function and its operands

Inspiration

• JavaScript: Destructuring
• CSharp: Tuples
• APL: short-left, long-right precedence

Example

The following code implements a Conway's Game of Life and updates until the count of active cells becomes unchanging

{
    print:Console/WriteLine
    Life:class {
        #@dbg["width:" w]
        w->i4
        #@dbg["height:" h]
        h->i4
        grid->Grid(bit)
        mod(n:i4 max:i4):(%: (<: n 0) ?+ (+:n max) ?- n max)
        # xy: (x:i4 y:i4)
        # %xy
        at(x:i4 y:i4):grid(mod.|[x:w y:h]|i4)
        get(x:i4 y:i4):at(x y)/Get()
        set(x:i4 y:i4 b:bit):at(x y)/Set(b)
        new(w:i4 h:i4): Life {
            print*cat["args: " _0 ", " _1],
            (w h) := ^^/(w h),
            grid := Grid(bit)/ctor(w h)
            debug()
        }
        debug(): print*|cat*|[["width: " w],["height: " h]]
        activeNum:0
        txt:StrBuild/ctor()
        update(): {
            activeNum := 0
            g:get
            txt/Clear()
            ta: txt/Append,
            ɩh | ?(y:i4){
                ɩw | ?(x:i4){
                    w:(-:x 1) n:(+:y 1) e:(+:x 1) s:(-:y 1)
                    c:count(g.|[w:n x:n e:n w:y e:y w:s x:s e:s] ⊤)
                    active: g(x y) ?+ not((<:c 2)∨(>:c 3)) ?- eq(c 3)
                    set(x y active)
                    active ?+ { activeNum := (+:_ 1) }
                    ta(active ?+ "*" ?- " ")
                }
                ta(newline)
            }
            print*cat["active: " activeNum]
        }
    }
    main(args:str) -> i4: {
        life:Life/new(32 32)
        print*life/grid[:i4 0 0]/Get(),
        ɩ(life/w) | ?(x:i4)ɩ(life/h) | ?(y:i4) life/set(x y rand_bool())
        num:1 prevNum:0 run:⊤ i:1
        Console/Clear()
        run ?++ {
            life/update()
            prevNum := num
            num := life/activeNum
            run := neq(num prevNum)
            print*cat["time: " i]
            Console/{
                SetCursorPosition(0 0)
                Write*life/txt/ToString()
            }
        }
        ^: 0
    }
}

Syntax

Oblivia has 3 basic structures.

• Array: Contains a sequence of items and nothing more.
• Tuple: Contains a sequence of items, some with string keys.
• Block: Contains a set of variables with string keys. Supports advanced operations such as ret

Arithmetic

Infix arithmetic is available for common operations:

• A + B
• A - B
• A × B
• A ÷ B
• A > B
• A < B

Lisp-like arithmetic allows you to spread operands. Operators are converted to reductions e.g. [+: a b c] = a/\+(b)/\+(c) = reduce([a b c] ?(a b) a/\+(b))

• [+: a b]
• [-: a b]
• [*: a b]
• [**: a b]
• [/: a b]
• [//: a b]
• [^: a b]
• [%: a b]
• [=: a b]
• [>: a b]
• [<: a b]
• [~: a b]
• [>>: a b]
• [<<: a b]
• [&: a b]
• [|: a b]
• [&&: a b]
• [||: a b]

Define

• A:B: field A has value B. If B is a type, then the value is a placeholder
• A -> B: Declare field A with type B
• A() -> B: Declare method A has type B`
• A -> B: C:
• A() -> B: C:
• A!:B: function A with no args has output B
• A(B, C): D: function A with args B,C has output D
• A[B C]: D:
• A{B C}: D:

Assign

• A := B: reassign field A to B (same type). You can use _ for the current value of A
• ^: A: Return A from the current scope.
• ^^: A: Return A from the parent scope.
• ^^^: A: Return A from the parent's parent scope.

Invoke

• A! = A(): call A with 0 args
• A*B = A(B): call A with arg B (associative-right)
• A.B = A(B): call A with arg B (associative-left)
• A(B C): call A with args B, C
• A.B.C.D = ((A*B)*C)*D = ((A(B))(C))(D)
• A*B*C*D = A(B(C(D)))
• A(B): If A is a generic type, then A(B) is the fully parametrized version of the type. If A is a non-generic or fully parametrized (e.g. does not accept generic arguments) type, then calling it simply casts B to that type.

Expression

• A: Get value of identifier A from the latest scope that defines it (current scope, then parent scope, then parent-parent scope)
• ^A: Get value of symbol A from the current scope
• ^^A: Get value of symbol A from the parent scope.
• ^^^A: Get value of symbol A from parent's parent scope.
• 'A: Alias of expression A. Assignments on variable B:'A will attempt to assign to A.
• [A B C]: Make an object array
• [A:B C:D] = [(A B), (C D)]
• [:type A B C]: Make an array of type
• { A }: Creates an scope and applies the statements A to it. If the scope has no locals or returns, then the scope returns the result of the last statement (empty if no statements). Otherwise returns an object.
• A ?+ B ?- C: If A then B else C.
• A ?++ B ?+- C ?-- D: While A, evaluate B. If at least one iter, eval C. If no iter, eval D
• A(B)
• A[B]: A([B])
• A{B}: A({B}) Call A with the result of {B}
  • If A is a class, then constructs an instance of A and applies the statements B to it.
• A-B: Range from A to B
• A->B: Function type
• A..B: Eval A, assign to _, then eval B
• A.B: A(B) Call A with arg term B (no spread)
• A*B: A(B) Call A with arg expression B (spread if tuple)
• A/B: In the scope of expression A evaluate expression B. Cannot access outer scopes.
• A/{B}: In the scope of expression A, evaluate statements B. Can access outer scopes.
• A/ctor: From .NET type A get the unnamed constructor.
• A|B: Map array A by function B.
• A?|B: Filter A by B
• ?(): A: Creates a lambda with no arguments and output A
• ?(A): B: Creates a lambda with arguments A and output B
• A.|B: B|A Call A with every item from term B (no spread)
• A*|B: B|A Call A with every item from expression B (spread if tuple)
• A/|B: ?(C) B(A C)
• A/|B(C): B(A C)
• A|.B: A|?(a):a(B) From every item in A call with arg term B
• A|*B: A|?(a):a(B) From every item in A call with arg expression B
• A|/B: A | ?(a) a/B From every item in A get value of symbol B.
• A||B: ?(C) A | ?(a) B(a C)
• A||B(C): A | ?(a) B(a C)
• A ?[ B0:C0 B1:C1 B2:C2 B3:C3 ]: Conditional sequence; for each pair B:C, if A(B) is true then include C in the result.
• A ?{ B0:C0 B1:C1 B2:C2 B3:C3 D0 D1 D3 }: Match expression (naive): For each pair B:C, if A = B, then returns C. Can also accept lambda D
• ?{ A0:B0 }: Matcher function
• A =+ B: Returns true if A equals B
• A =- B: Returns true if A does not equal B
• A = B: Returns true if A matches pattern B
• A = B:C: Returns true if A matches pattern B and assigns the value to C
• A =: B: if A = typeof(B), then sets A := B and returns true

Pattern

• $(A:B)
• $[A B C]: Array
• $[A:B C:D]:
• $[A=B C=D]:
• ${ A = B }: Object member A of type B
• ${ A = B:C }: Object member A of type B; make local C:B(A)
• ${ A:B }: Object member A of type B; define A:B

Constants

• yes: True
• no: False
• ⟙: True
• ⟘: False
• ∅: Empty
• empty: Automatically removed when added to a tuple or array

Monadic functions

• ↕A: Returns [0 1 2 ... A]
• ⍋A: Returns [n n+1 n+2 ... m] where [A(n) A(n+1) A(n+2) ... A(m)] = sorted(A)
• ⍒A: Returns [n n+1 n+2 ... m] where [A(m) ... A(n+2) A(n+1) A(n)] = sorted(A)
• ⌈A: ceil(A)
• ⌊A: floor(A)
• A⋖: First element of A
• A⋗: Last element of A
• A⌗: Returns length of A
• ⚄A: Returns a random item from [0 1 2 ... A]
• ⌨A: Returns the value of the first char of A
• ¬A: Returns not(A)

Dyadic functions

• A⌗B: Returns count of B in A
• A∀B: Returns A ?| B⌗ = A⌗
• A∃B: Returns A ?| B⌗ > 0
• A∄B: Returns A |? B⌗ = 0
• A⫷B: Constructs class A with data B
• A∨B: Returns [||:A B]
• A∧B: Returns [&&:A B]
• A⋃B: Returns any(A B)
• A⋂B: Returns all(A B)
• A≤B:
• A≥B:

Design philosophy

• Whitespace is the simplest operator.
  • Two adjacent identifiers A B simply means that A occurs before B in a sequence. Identifiers are never grouped together outside of tuples and arrays. { A B:C D } = { A:A, B:C, D:D }, { (A B):(C D) } = { A:C, B:D }
  • There is never a statement of the form A B such that A,B are identifiers and A performs some operation on B, other than occurring earlier in a sequence.
• Function calls with 0/1 arguments are allowed alternate syntax to save pixels
  • Calls with n>1 arguments always require enclosing delimiters A(B C)
  • Calls with 0, 1 arguments are allowed single-ended operators A!, A*B, A.B

Rejected features.

OBL emphasizes generality and terseness, rejecting features that are not versatile enough to justify the syntax cost.

• Partial application / whatever-priming (Raku): Scope-constrained to simple expressions. Cannot control argument order.
  • ?(<par>) <expr> Lambdas solve this problem by forward declaring arguments and allowing any size scope.
• =-based assignment: This function is often confused between assignments and boolean comparisons.
  • OBL uses :=, where : makes it clear that this function is strictly assignment. = is a pattern match function.