Publish Quarto Dashboard

Prompt: Is there a way to use shinylive to publish demo.qmd (demo.html) as a static file much as we did with the shiny app?

Response: No, you cannot directly export a standard server: shiny Quarto Dashboard (which uses markdown sections for the layout and separates the server code in #| context: server blocks) using shinylive::export.

Here is why, and how you can achieve a static Shinylive version of the dashboard. [The advice would work for a much simpler app. It did not work in this case.]

Why a standard Shiny Quarto Dashboard cannot be exported directly

Server dependency: Standard Quarto dashboards with server: shiny rely on a running backend R session (like Posit Connect or Shiny Server) to sync the inputs/outputs spread throughout the document.
Shinylive’s “Standalone” requirement: When using Shinylive in Quarto, the entire Shiny application (both UI and Server logic) must be self-contained inside a single code block defined as #| standalone: true. You cannot distribute widgets across different markdown headings/sidebar columns.

How to publish it statically with Shinylive

If you want a serverless, static version of this dashboard, you have two options:

Option A: Use the R Shiny App (Already Done)

The Shiny application in inst/connect_modules/app.R already contains the same plots and logic as demo.qmd.

Because it is a standard Shiny app, it is fully compatible with shinylive::export().
It is already successfully exported as a serverless, static app in your docs/shiny/ directory.

Option B: Embed a Shinylive App inside a Quarto Document

If you specifically want a .qmd document that compiles to a static page containing the interactive dashboard, you can embed the R code inside a {shinylive-r} block:

Install the Shinylive Quarto Extension (locally in your terminal):
```
quarto add quarto-ext/shinylive
```

Create a new .qmd file (e.g. docs/demo_shinylive.qmd) with the shinylive filter:

---
title: "Geyser Dashboard (Static)"
format: html
filters:
  - shinylive
---

Embed the code in a single standalone block:

```{shinylive-r}
#| standalone: true
#| viewerHeight: 700

# Paste the loading helper, ui structure, and server logic here
# (e.g. your source code from inst/connect_modules/app.R)

```

Render it: When you run quarto render docs/demo_shinylive.qmd, it will compile into a serverless interactive HTML page that runs entirely in the browser and can be served statically on GitHub Pages.

Walkthrough: Option B Implementation

I have implemented and automated Option B to host a serverless, interactive Quarto dashboard on GitHub Pages:

1. Installed Quarto Shinylive Extension

Installed the extension in the repository:

quarto add quarto-ext/shinylive --no-prompt

2. Created Self-Contained Shinylive Document

Created docs/demo_shinylive.qmd. It is structured to run completely in the browser via WebAssembly:

YAML Header: Configured format as html and registered the shinylive filter. Explicitly specified engine: knitr to bypass Jupyter/Python kernel execution checks (avoiding errors like ModuleNotFoundError: No module named 'nbformat').
Code Chunk: Set #| standalone: true and concatenated all the R modules (from the root R/ directory) and the navbar UI layout (from app.R) inside a single {shinylive-r} code block.

3. Integrated into CI/CD Deployment Workflow

Updated .github/workflows/deploy.yml to automate the build-and-deploy process on GitHub Actions:

Installs shinylive, knitr, and rmarkdown R packages inside the runner.
Installs the Quarto Shinylive extension inside the GitHub Actions runner.
Renders docs/demo_shinylive.qmd using quarto render docs/demo_shinylive.qmd.
Deploys the output docs/demo_shinylive.html (along with the other pages in the docs/ folder) directly to GitHub Pages.

This avoids local rendering Deno cache errors, bypasses unnecessary Jupyter python dependencies, and automates updates every time you push to the main or master branch.