Don’t let long running tasks hang users: introducing IPC for Shiny
summary
Long-running Shiny computations run in separate R processes using ipc package queues and the future backend so the server and other outputs stay responsive. The system sends eval/call messages (file-backed by txtq or Redis) and provides a drop-in async progress and an async interrupter to show progress and cancel futures; examples include a 1..100 counter and a color-cycling histogram and they work with minimal file I/O overhead.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:13Design considerations
- 2:30Hello world example
- 4:03Asynchronous computation
- 6:45Good and bad
- 7:08What is IPC
- 8:49IPC methods
- 11:36Shiny features
- 12:52Examples
related talks
32mAchieving Concurrency in Streamlit with a RQ scheduler
35mBuild Simple and Scalable Apps with Shiny
44mMaking apps seem faster without optimizing
19mDisposable Shiny Apps
27mShiny: A GUI for the Tidyverse
44mShiny: Data-centric web applications in Python
01h 00mInteractive Computing with Jupyter: Past, Present, and Future
38mProcrastination as a Service: Caching and Queueing