Stenobird

Episode

#478 Iodine tablets and potable water

Podcast
Python Bytes
Published
May 4, 2026
Duration seconds
2413
Processing state
processed
Canonical source
https://pythonbytes.fm/episodes/show/478/iodine-tablets-and-potable-water
Audio
https://pythonbytes.fm/episodes/download/478/iodine-tablets-and-potable-water.mp3
JSON
/v1/public/podcasts/python-bytes/episodes/478-iodine-tablets-and-potable-water
Markdown
/podcast/python-bytes/478-iodine-tablets-and-potable-water.md

Actions

  • POST https://stenobird.com/v1/public/podcasts/python-bytes/episodes/478-iodine-tablets-and-potable-water/transcription-requests
    Idempotently request low-priority transcript generation for this episode.
  • GET https://stenobird.com/podcast/python-bytes/478-iodine-tablets-and-potable-water.md
    Read the agent-friendly Markdown representation of this episode resource.

Summary

Python developers face a major regression as the core team reverts the incremental garbage collector in 3.14 and 3.15 due to extreme memory pressure. The episode also explores new tools for profiling and static site generation for Django.

Topics

  • Python
  • Garbage Collection
  • Django
  • Software Profiling
  • Static Site Generation
  • AI Coding Assistants
  • DevOps
  • Software Performance

Highlights

  • Main idea: The incremental GC in Python 3.14/3.15 is being reverted to the generational model to prevent massive memory spikes in cyclic workloads
  • Practical takeaway: Use profiling-explorer to visually navigate pstats files and identify performance bottlenecks via searchable, sortable tables
  • Practical takeaway: Django Freeze offers a low-risk way to deploy Django apps as static sites, reducing DevOps complexity and operational overhead
  • Failure mode: The new incremental GC achieved low latency (1.3ms) but caused a 5x increase in peak RSS memory usage
  • Main idea: AI-driven development tools like VS Code's Co-author are shifting the focus from simple autocomplete to complex agentic workflows

Chapters

  1. 1:00 Introducing profiling-explorer: A look at Adam Johnson's new tool for exploring Python profiling data with a user-friendly, searchable interface.
  2. 4:05 The Python GC Revert: Discussion on why the Python core team is reverting the incremental garbage collector in versions 3.14 and 3.15.
  3. 10:00 Django Freeze: Exploring a project that allows developers to convert Django applications into highly reliable static sites.
  4. 22:05 AI Co-authors and VS Code: The impact of AI agents and the debate surrounding AI-integrated development environments.
  5. 34:30 GitHub Stability and Infrastructure: Reflections on recent GitHub service instability and the importance of reliable developer infrastructure.
  6. 37:20 The AI Apocalypse Joke: A humorous look at the cultural divide between tech enthusiasts and those wary of AI's rapid advancement.