Logs, Tears and Feature Flags: Code Stories in Survival Mode

on YouTube ↗15m

" summary

A Python app rewrite using Streamlit stores critical sleep-and-feeding records in a local JSON store instead of Streamlit session state so data survives crashes and different devices. Nonessential charts and emojis are gated by LaunchDarkly feature flags, and a runtime "survival kill switch" context disables optional features and an unreliable external API without redeploy. The setup preserves offline critical data while letting experiments be toggled live so the app remains usable at 3 a.m.

AI-written summary · about

" chapters (9)

1. 0:00Tech failure night
2. 1:55Dev to mom
3. 3:39Python app with Streamlit
4. 5:58Architectural error
5. 7:22Adding charts and emojis
6. 9:12External dependency risks
7. 11:23Feature flags saved production
8. 13:28UI, data and recovery
9. 14:10Maternity and dev lessons

" related talks

• 29mSelf-Destructing Feature FlagsJamie Gaskins · SREcon22 Americas
• 54mFeature Flags Separate Deployment from Feature ReleaseRob Richardson · ConFoo 2025
• 42mStreamlit: A Faster Way to Build and Share Data AppsKaren Javadyan · PyData Yerevan 2022
• 28mBeyond Prototyping: Building Production-Level Apps with StreamlitJohannes Rieke, Arnaud Miribel · PyData Paris 2025
• 01h 58mHow to create beautiful interactive GUIs and web appsLisa Carpenter · PyCon US 2023
• 01h 09mTurn Python Scripts into Beautiful ML ToolsAdrien Treuille · PyData Los Angeles 2019
• 15mFeature FlagsRaven Covington · Bath Ruby 2018
• 18mStreamlit: The Fastest Way to build Data AppsSteven Kolawole · Conf42 Python 2021