Why XPR

Existing expression languages fall short for cross-language data pipeline use cases — they're either tied to a single runtime, lack modern collection operations, or carry heavy dependencies. XPR fills that gap.

How it Compares

	JEXL	CEL	JSONata	Starlark	Expr (Go)	XPR
Lambdas / arrow functions	❌	❌	Limited	✅	✅	✅
Collection methods (map, filter, etc.)	❌	Limited	✅	✅	Limited	✅
Cross-runtime (JS + Python + Go)	Partial	❌	❌	❌	❌	✅
JS/Python-familiar syntax	✅	❌	❌	Partial	❌	✅
Zero runtime dependencies	✅	❌	✅	❌	✅	✅
Shared conformance tests	❌	❌	❌	❌	❌	✅

Key Properties

• Cross-runtime parity — The same expression produces identical results in JavaScript, Python, and Go, verified by 600+ shared conformance tests across 22 test files.
• Familiar syntax — Arrow functions (x => x * 2), template literals, optional chaining, pipe operator (|>), destructuring. If you know JS or Python, you already know XPR syntax.
• Safe sandbox — No I/O, no side effects, no prototype access. Configurable execution timeout (default 100ms) and AST depth limit (default 50).
• Zero dependencies — Each runtime is a hand-written Pratt parser and tree-walk evaluator. No third-party parsing library, no generated code, no build step.
• Pure expressions — No statements, no mutation, no loops. Iteration is collection methods only.

When to Use XPR

• Dynamic business rules — Evaluate user-defined filter/transform rules against structured data at runtime.
• User-configurable pipelines — Let end-users write expressions that run safely inside your application.
• Multi-tenant formula evaluation — Execute untrusted expressions in a sandbox where host access is impossible.
• Cross-platform portability — Author expressions once and evaluate them identically in a JS frontend, Python backend, and Go service.

When NOT to Use XPR

• General-purpose programming — XPR has no while, no class, no I/O, no imports. Use a real scripting language.
• Stateful computation — let bindings are immutable; there is no variable reassignment or persistent state across evaluations.
• Native performance requirements — XPR is a tree-walk interpreter. For tight loops over large datasets, use native code.
• When you need host access — Expressions cannot reach the filesystem, network, or environment variables by design.