Why code autocompletion works faster on weekends
summary
Production opt in telemetry shows autocompletion faster on weekends because users work on smaller projects then and completion latency correlates with project symbol count, not server load. Benchmarks catch regressions but miss real-world variability. A developer survey about the old native memory model yields 60% mistaken responses on core examples. A new ranges operator cuts errors and time — up to 25% improvement in the most common for-loop scenario — and a native garbage collector removes many surprising behaviors.
AI-written summary · about
chapters (6)
- 0:00Kotlin team and IDE performance
- 3:17Kotlin Native memory model issues
- 8:40Range declaration in Kotlin
- 13:40API design lessons
- 14:44Autocompletion faster on weekends
- 16:50Q&A
related talks
26mFrom autocomplete to agents: AI coding assistance state of play
29mBeyond Autocomplete: Local AI Code Completion Demystified
01h 16mHow METR measures Long Tasks and Experienced Open Source Dev Productivity
02h 31mDeveloper Productivity Engineering - Extended Session
49mCopilot, Cursor, and Custom LLMs: Navigating the New .NET Developer Experience
18mFast Models Need Slow Developers
37mCrafting Faster in the Dark
01h 21mSoftware Development - Mob Programming, Pair Programming or Fly Solo?