I present Time Series App Cookbook — a runnable Dash template for navigating and annotating recordings of millions of points without lag, built on plotly-resampler. The demo makes 2.7M synthetic samples, so you can feel the speed. The README is the cookbook: 18 recipes, each saying what a feature does, how it works, and what breaks if you change it.
A while ago I posted a toy Dash app for time series labeling. This cookbook is that toy grown up — loaded with useful new features: drag-to-select with auto-pan, one-step undo and crash recovery, multiple independent windows, native file dialogs. Both are distilled from the same parent: a sleep-scoring app I built, which the labs in my research program use daily to score mouse EEG/EMG/NE recordings.
Desktop App Source Updater — how we ship it. PyInstaller freezes only a launcher; the app’s own .py files sit beside the bundle and update themselves from GitHub Releases at startup. A bugfix is a small zip, not a full rebuild.