Making an App a System
summary
Turns a canonical Shiny app into a versioned R package that child apps depend on, moves heavy data wrangling out of Shiny into scheduled jobs, and uses simple YAML config files per child to map differently named columns and settings. Uses Posit Connect scheduled jobs to produce cached datasets and demonstrates the minimal proof-of-concept {matte} package on Atorus GitHub that scaffolds a template app, generates child repos, and successfully reduces duplication and startup cost while staying in early development.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:44Pharma and GxP constraints
- 2:13Copying apps for studies
- 3:39Package-based system
- 4:47Separating data processing
- 5:52Configuring with YAML
- 6:55Versioning and validation benefits
- 7:41Introducing matte package
- 8:47Generalized strategy recap
- 9:15Q&A on caching
- 10:21App code in packages
- 11:07Shiny frameworks: Golem vs Rhino
related talks
13mIntroducing Rhino: Shiny application framework for enterprise- 13mIntroducing Rhino: Shiny application framework for enterprise
18mModular, layout-as-code approach for customizable Shiny dashboards
49mScalable Best Practices for Your Development Team
07mStreamlining Development with the Monorepo Ecosystem
21mShared Components Beyond the Design System
05mLeveraging CSS to Make Shiny Mobile-Responsive
22mShort story: When a close-knit JS App family needed rescue from an unruly State butler