LibPolyCall
A C runtime broker that lets Python, Node.js, Java, Go and even COBOL code call each other's functions through one unified, program-first protocol instead of writing custom bridges for each language pair.
🔗 Visit LibPolyCall
Description
Teams building systems that mix several programming languages — a Python data pipeline talking to a Java backend, or a modern service that still needs to call into old COBOL code — usually end up writing a custom integration layer for every single pair of languages involved. LibPolyCall tries to replace all of those one-off bridges with a single broker: you define your API once, and any supported language can call it, without needing to know or care what language wrote the other side.
Technically, LibPolyCall is a C-based runtime broker with official bindings for Python, Node.js, JavaScript, Java, Go, and COBOL, built around a 'program-first' polymorphic protocol design and a zero-trust security model with cryptographically-seeded GUIDs for tracking state and telemetry across calls. It's aimed at enterprise-style polyglot codebases and legacy integration work — the COBOL support in particular signals it's meant for organizations still running mainframe-era systems alongside modern services. It's actively developed (231 commits, v1.1.0), dual-licensed, and documented with a fairly comprehensive README covering configuration and migration guidance.
💬 Our review
The short version: LibPolyCall is a bet that one well-designed cross-language protocol beats writing a custom FFI/RPC bridge for every language pair in a polyglot stack — and its COBOL support is the detail that makes that bet interesting for legacy-heavy enterprises.
Compared to established alternatives — gRPC, Thrift, or hand-rolled FFI — the pitch is architectural elegance: define once, call from anywhere, rather than maintaining N-squared bridges between languages. That's a real problem worth solving, especially where COBOL is involved, since almost nothing modern bothers to support it. The catch is adoption: at 4 GitHub stars and 0 forks, there's essentially no independent validation that the protocol holds up under real production load, and the ambitious security/telemetry claims in the README haven't been tested by any outside team as far as the public repo shows. It's free and open source, so the cost of trying it is just engineering time. Worth a serious look if you're specifically stuck bridging COBOL or another odd legacy language into a modern stack and existing options don't cover it; for anything gRPC or Thrift already handle well, those remain the safer, battle-tested choice for now.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Gratuit, double licence
Pros
Protocole unifié pour appeler du code Python, Node.js, Java, Go et COBOL sans bridge dédié par paire de langages
Support COBOL rare parmi les outils modernes d'interopérabilité
Développement actif (231 commits) et documentation détaillée
Cons
Adoption quasi nulle (4 étoiles GitHub, 0 fork), aucune validation indépendante en production
Positionnement expérimental, maturité réelle incertaine
Alternatives établies (gRPC, Thrift) restent le choix par défaut hors cas COBOL