70,000 Lines of Code, 7 Languages: How to Scale via Rust Core @infoq
70,000 Lines of Code, 7 Languages: How to Scale via Rust Core  @infoq
Uploaded July 2026 | Updated November 2026, 1 day ago
The multi-language SDK trap: Writing the same complex state machine seven times in seven different languages is a recipe for architectural drift, bugs, and developer burnout.

In this InfoQ video, Spencer Judge (SDK Team Lead at Temporal) shares the exact blueprint Temporal used to unify their heavy client-side logic into a shared, lightning-fast Rust core wrapped in language-specific bridges.

Learn the hard-won architectural lessons of cross-language development, from handling memory management across the C FFI boundary and syncing async runtimes (Promises vs. Futures vs. Fibers), to why WebAssembly and zero-serialization IDLs like FlatBuffers might be the future of your shared codebase.

⏱️ Video Timestamps (For Navigation)
0:00 - The Problem: 70,000 Lines of Code across 7 Languages
01:45 - Why Client-Side Logic Needs to be "Fat"
02:50 - Architecture Breakdown: The Rust Core and Bridge Layer
04:15 - Deep Dive into Bridge Helpers (PyO3, Neon, Magnus)
05:40 - The Type Conversion Nightmare & Macro Solutions
07:10 - Pop Quiz: Is Direct Memory Move Faster than JSON Serialization?
08:50 - Bridging Async: Rust Futures vs. Python/JS Promises vs. Ruby Fibers
10:30 - Unsafe Rust & Cross-Boundary Memory Management Rules
12:10 - Structural Limitations & Shipping Native Code (npm, Python, Go)
13:45 - What We'd Do Differently: WebAssembly (Wasm) and Pure VMs (Wazero)
15:30 - Zero-Serialization IDLs: Cap'n Proto & FlatBuffers
16:40 - Final Verdict: Was the Rust Core Architecture Worth It?

πŸ”— Transcript available on InfoQ: bit.ly/4w4oQ4N

#SoftwareArchitecture #RustLang #Temporal #WebAssembly #DistributedSystems
70,000 Lines of Code, 7 Languages: How to Scale via Rust CoreHow Netflix Built a Single Model for Search & RecommendationsWhat Happens When Your Entire Cloud Provider Fails?  #infoq20 years of InfoQ πŸŽ‚ #MichaelRedlich #InfoQ20 #DevCommunityThe Reckless Speed of AI Agents: Architectural Amnesia EXPOSEDHow I Doubled eBay’s Productivity - And Still Got FiredChris Richardson EXPOSES the Microservices Cognitive Load Trap20 years of InfoQ πŸŽ‚ #ClaudioMasolo #InfoQ20 #DevCommunityCloud and DevOps InfoQ Trends Report 2026: AI, Resilience, Platforms, FinOps, and SovereigntyHow Meta Scaling AI-Native Engineering Tripled Code Coverage (And Cut Slop)RAG: Similarity β‰  RelevanceAI isn’t your bottleneck. Code review isπŸ›‘  #infoq
InfoQ |

70,000 Lines of Code, 7 Languages: How to Scale via Rust Core

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER