GreffonExpression trees for TypeScript

Write an ordinary lambda; it stays the function it always was, and becomes a typed, serializable tree you can evaluate, rewrite, print, store, send over the wire — or hand to a provider that translates it: a policy check, a remote filter, parameterized SQL.

Get started

Try the playground

Applications

The tree is the product

A typed, versioned, JSON-serializable algebra with zero dependencies and a public schema. Evaluate it without its function, rewrite it, print it, ship it across a process boundary.

The expression tree

One lambda, two lives

The same query file runs against fixture arrays in your tests and compiles to a parameterized SQL WHERE clause in production. The in-memory provider is the reference semantics; the SQL provider is property-tested against it.

Getting started

Policy checks, wire filters, rules as data

The predicate that filters a list query also authorizes one object, runs in the browser, and serializes into an auditable store — one definition, four jobs.

Applications

One shared validator

The subset validator, free-variable analysis, and serializer live once and back the build, the editor, the ESLint rule, and the runtime fallback — so they never disagree about what is legal.

The expression subset

Fail fast, never guess

Untranslatable queries are located, coded errors — not silent client-side table scans. Rows cross into JavaScript only at an explicit inMemory() boundary.

The boundary rule

Next to your stack

Greffon is the query layer, beside whatever owns your schema and writes. Here is how it relates to Prisma, Drizzle, Kysely, TypeORM, MikroORM, EF Core, and the rules engines.

Compared to ORMs & rules engines