Skip to main content

​
Value Types in Stoffel

This page explains the value types and data structures available in Stoffel’s virtual machine, their properties, and how they’re managed during program execution.

​
Rationale

Stoffel’s type system is designed to provide the fundamental building blocks needed for practical programming. By implementing a carefully selected set of primitive and composite types, Stoffel enables developers to create complex data structures and algorithms while maintaining consistent behavior across different execution contexts.

​
Comprehensive Overview

Stoffel’s type system includes primitive values (integers, floats, booleans, strings), composite values (objects, arrays), and function values (closures). Each value is represented by the Value enum, which encapsulates both the type and data of a value. The VM maintains separate storage mechanisms for different composite types. Objects and arrays are managed through the ObjectStore, which assigns unique identifiers to each instance. This reference-based approach allows for efficient manipulation of complex data structures while maintaining proper isolation between different parts of a program.

​
Primitive Types

  • Int(i64): Represents integer values.
    • Used for counting, indexing, and whole-number arithmetic
    • Supports all standard arithmetic and bitwise operations
  • Float(i64): Represents floating-point values using fixed-point representation.
    • Stored as fixed-point (scaled integers) for consistent behavior
    • Maintains equality and hashability
    • Display format converts to standard floating-point notation
  • Bool(bool): Represents boolean values.
    • Used for logical operations and conditional expressions
    • Values are either true or false
  • String(String): Represents text data.
    • Used for identifiers, keys, and textual content
    • Immutable once created
  • Unit: Represents the absence of a value.
    • Similar to null or void in other languages
    • Used as default return value for functions with no explicit return

​
Composite Types

  • Object(usize): Represents a key-value mapping structure.
    • Reference to an object in the VM’s object store
    • Fields can be accessed using any value type as keys
    • Supports dynamic addition and modification of fields
  • Array(usize): Represents an ordered collection of values.
    • Reference to an array in the VM’s array store
    • Optimized for integer-indexed access starting from 1
    • Supports both dense and sparse representations
    • Also allows non-integer keys for associative array behavior

​
Function Types

  • Closure(Arc<Closure>): Represents a function with its captured environment.
    • Contains a reference to a function and its upvalues
    • Enables higher-order functions and callbacks
    • Preserves lexical scoping through upvalue capture

​
Foreign Values

  • Foreign(usize): Represents a reference to an external object.
    • Enables integration with host environment resources
    • Type-safety maintained through dynamic type checking
    • Allows extension of the VM with custom types and operations

​
Arrays

Arrays in Stoffel are implemented as a hybrid data structure:
  • elements: A contiguous vector of values for efficient integer-indexed access
  • extra_fields: A hash map for sparse or non-integer indices
  • Integer indices start from 1 (following traditional scripting language conventions)
  • Arrays automatically grow when elements are added at the end
  • Sparse indices are handled efficiently to avoid wasting memory

​
Objects

Objects in Stoffel are implemented as general purpose key-value stores:
  • fields: A hash map supporting any value type as both keys and values
  • No predefined structure, allowing for dynamic addition and removal of fields
  • Enables flexible data modeling patterns, including record and dictionary styles

​
Object Store

The ObjectStore provides centralized management of objects and arrays:
  • Assigns unique numeric identifiers to each instance
  • Maintains separate collections for objects and arrays
  • Provides controlled access through getter methods
  • Handles field access and modification with appropriate type checking
  • Ensures proper isolation between different data instances

​
Foreign Object Storage

The ForeignObjectStorage enables safe integration with external types:
  • Type-erased storage through the AnyObject trait
  • Type-safe retrieval using generics and downcasting
  • Thread-safe access using Arc<Mutex<T>> for concurrent operations
  • Allows extension of the VM with custom native types

​
Upvalues and Closures

Closures in Stoffel capture their lexical environment through upvalues:
  • Upvalue: A named variable captured from an outer scope
  • Closure: A function combined with its captured upvalues
  • Upvalues maintain a deep copy of the captured values
  • Ensures predictable behavior across different execution contexts