miniKanren: Discussion on the Future of miniKanren
summary
miniKanren has ~17–18 years of history, dozens of tiny implementations and a few heavier ones; concrete wins include a relational interpreter that enables program synthesis and type inhabitation, the neural‑guided search (NIPS 2018) experiment, and applied systems like DB canren for biomedical graph queries and core.logic in industry. Current pain points are noncommutative conjunction ordering causing catastrophic performance, sparse modularity, fragmented constraint libraries (finite domains, nominal unification, SMT bridges), and missing garbage collection in substitutions. Future directions push for a modular mega‑Canren with pluggable searches, solver hybrids and richer data structures.
AI-written summary · about
chapters (10)
- 0:00Evolution of miniKanren
- 4:29Bridging gaps in knowledge
- 8:20Implementation diversity
- 12:59Strategic future directions
- 18:06Pragmatic use cases
- 25:33Defining the core technology
- 32:28Vision for a megaKanren
- 44:26Practical applications and needs
- 58:20Relational glue and systems
- 1:10:25Operational semantics and optimization
related talks
01h 30mTrustworthy AI & Open Source
21m#10 Open Source Forum 2019 Shintarou Okada
44mOpen collaborations: leadership succession and leadership success
15mThe CNCF Sandbox: An Exploration and Guided Tour
51mPanel Discussion
42mState of the Project & State of the Industry
06mProject Lightning Talk: Architecture Without Gatekeepers: Decoupling for Neutrality
22mDesign Automation in Wonderland: EPFL Logic Synthesis Libraries