Balli vs Malli

Balli is Malli for Basilisp: a data-driven schema library that ports Malli's schema syntax and core workflows into Python-hosted Basilisp code. It is not a byte-for-byte copy of Malli's Clojure implementation; the differences below are the places where Basilisp/Python runtime constraints or ecosystem maturity change the shape of the library.

What Balli Covers

Balli currently covers most user-facing schema workflows:

  • validation and explain data
  • humanized errors
  • coercion, decoding, encoding, parsing, and unparsing
  • non-throwing coercion reports and transformer roundtrip checks
  • map, collection, predicate, comparator, enum, ref, multi, and regex schemas
  • function schemas, instrumentation, and generative checking
  • deterministic generators, shrinking, shrink traces, and small property-check reports
  • Balli-native generator objects and combinators
  • JSON Schema, OpenAPI, Swagger, DOT, PlantUML, and description export
  • schema walking, utility operations, map-entry surgery, path rewrites, and diffs
  • custom registries, local registries, lazy/dynamic/var registries, and recursive references
  • EDN-style safe serialization through registered function references
  • custom extension types
  • time schemas
  • schema inference, provider fit reports, and destructuring helpers
  • data-only experimental analysis helpers, including dependency graphs, migration impact, and static risk reports
  • Python integration adapters, callable coercers, and static schema inspection

Remaining Differences

Some Malli ecosystem features are intentionally thinner in Balli:

  • Malli's ClojureScript/browser ecosystem has no direct Basilisp equivalent.
  • Malli's documentation site and community examples are more mature.
  • Malli has broader downstream library integrations and more community examples.
  • Malli's advanced provider ecosystem and schema tooling are still deeper, but Balli now has provider knobs, fit reports, static inspection, path diffs, and data-only experimental analysis helpers.
  • Balli's package ecosystem is younger because Basilisp's ecosystem is smaller.

Practical Parity Target

The most important parity target is not copying every implementation detail. It is making the common production path solid:

  • install from PyPI
  • learn from hosted docs and examples
  • validate and transform data at API/config boundaries
  • export OpenAPI/JSON Schema
  • run a fast user suite after cloning
  • rely on CI, fuzz tests, stress tests, and release checks

That path is now present. The remaining parity work is mostly ecosystem depth: more tutorials, more integrations, more polished docs, and a custom domain if the project becomes public-facing enough to justify it.