_05_shiny_in_prod.Rmd

---
title: "Shiny in prod"
output:
  xaringan::moon_reader:
    css: [default, rd.css, rd-font.css, "hygge"]
    lib_dir: assets
    seal: false
    nature:
      highlightStyle: dracula
      countIncrementalSlides: false
      beforeInit: "macros.js"
---

```{r include = FALSE}
source("_setup.R")
```

class: inverse, center, middle

# `r rfa("rocket")` Shiny in production

![:custom_hr]()

## An overview


---
# Deploying a Shiny app 

## Making your application available 

<br>

--

* Posit (formerly RStudio) solutions:
    * `r lk()` https://posit.co/products/open-source/shinyserver/
    * `r lc()` https://www.shinyapps.io/
    * `r lk()` https://docs.posit.co/shinyapps.io/index.html
--

* Self-hosting:
    - `r rfa("blog")` https://emeraldreverie.org/2019/11/17/self-hosting-shiny-notes-from-edinbr/
    - `r lk()` https://www.shinyproxy.io/#open-source

???
Account via github

---
# Best practices

### `r bk()` https://mastering-shiny.org/scaling-intro.html

--

## Follow general best practices 
### (software engineering principles)

* Good design of your code (easy to read, non-redundant, etc.)
* version control such (e.g. `r gt()`)
* use a good code editor
* test your application
* etc.


---
# Best practices

## Pay attention to your `r rp()` code

* Use functions to avoid redundancy
* Use mapping functions [`purrr`](https://CRAN.R-project.org/package=purrr)

--

## Profile and optimize your `r rp()` code

* profile your code
  * `Rprof()`
  * `r rfa("box")` [`profvis`](https://CRAN.R-project.org/package=profvis) ([flame graph](https://queue.acm.org/detail.cfm?id=2927301))
* benchmark
  * `r rfa("box")` [`microbenchmark`](https://CRAN.R-project.org/package=microbenchmark)
* read about this 
  * `r rfa("bookmark")` [Improving performance](https://adv-r.hadley.nz/perf-improve.html) (`r bk()` Advanced R) 
  * `r bk()` [Efficient R programming](https://csgillespie.github.io/efficientR/)


---
# Best practices

### **More complex shiny longer codebase**

--

### Try to create **independent components**

* functions in separate `.R` files in `R/`
* external package(s)
* modules


---
# Best practices


### Use [modules](https://mastering-shiny.org/scaling-modules.html) 

```R
yourModuleUI <- function(id) {
  ns <- NS(id)
  tagList(...)
}

yourModuleServer <- function(id, geoms, preview, u_details) {
  moduleServer(
    id,
    function(input, output, session) {
      ...
    }
  )
}   
```




---
# Best practices

## Turn your shiny app into an `r rp()` package 

### `r ar()` Access to robust tools to develop your app (namely `r rfa("box")` [`devtools`](https://CRAN.R-project.org/package=devtools))

--

### **Testing**

* Critical part of the development of your app
* Unit tests for your functions (see `r rfa("bookmark")` https://r-pkgs.org/testing-basics.html#introducing-testthat)
* Test reactivity with `testServer()`(see `r rfa("bookmark")` https://mastering-shiny.org/scaling-testing.html)
* Test your application visual (using snapshot, see https://testthat.r-lib.org/articles/snapshotting.html)


---
# Best practices

### Avoid heavy computations when possible

--

### Load tests

  * Test how your application with a given load (i.e. a given number of users using your application simultaneously)
  * `r rfa("box")` [`shinyloadtest`](https://CRAN.R-project.org/package=shinyloadtest)

--

### Caching 

  * Save the output result of a given action and reuse it instead of redoing the computation
  * Use `bindCache()` (see also `r rfa("box")` [`cachem`](https://CRAN.R-project.org/package=cachem))
  * `r lk()` https://shiny.rstudio.com/articles/caching.html


--

* `r rfa("youtube")` https://www.rstudio.com/resources/rstudioconf-2019/shiny-in-production-principles-practices-and-tools/