Reproducible workflow

---

# Think of your R processes as livestock, not pets

---

## Pets or cattle?

---

## R Session

* R process (e.g. "session")
* Treat individual R processes and workspaces as disposable

### Workspace

* Libraries with `library()`
* User-created objects

---

---

## Save code, not workspace

* Enforces reproducibility
* Easy to regenerate on-demand
* Always save commands
* Always start R with a blank state
* Restart R often

---

## Always start R with a blank slate

---

## Bad approaches

```r
rm(list = ls())
```

* Good intent, but poor execution
* Only deletes user-created objects
* Enforces hidden dependencies on things you ran before `rm(list = ls())`

---

## Avoid unknown unknowns

Write every script like its running in a fresh process

Best way to ensure this: **write every script in a fresh process**

### Storing computationally demanding output

* `write_rds()` & `read_rds()`
* `cache: true`

---

# Working directories and filepaths

---

## How to store work

* Split work into projects
* **We already do this**
* But why?

---

## Working directory

- Directory in a hierarchical file system dynamically associated with a process
- `getwd()` and `setwd()`

## `setwd()`

```r
library(tidyverse)
setwd("/Users/bensoltoff/cuddly_broccoli/verbose_funicular/foofy/data")
foofy <- read_csv("raw_foofy_data.csv")
p <- ggplot(foofy, aes(x, y)) + geom_point()
ggsave("../figs/foofy_scatterplot.png")
```

---

## Relative and absolute paths

#### Relative path

```
data_world_bank/API_ABW_DS2_en_csv_v2_4346306.csv
```

#### Absolute path

```
/Users/soltoffbc/Projects/Computing for Information Sciences/homework-seeds/hw04/data_world_bank/API_ABW_DS2_en_csv_v2_4346306.csv
```

Absolute paths will not work for anyone besides the original author - and even for them they will eventually break

**Use relative filepaths**

---

# Project-based workflows

---

### File system discipline

Put all files related to a single project in a designated folder

### Working directory intentionality

When working on project A, make sure working directory is set to project A's folder

### File path discipline

All paths are relative - relative to the project's folder

### Rationale for workflow

* Ensures portability
* Reliable, polite behavior

### RStudio Projects

* `.Rproj`

---

## Use safe filepaths

* Avoid `setwd()`
* Split work into projects
* Declare each folder as a project
* Use `here::here()`

---

## `here::here()`

```r
library(here)
here()
```

```
## [1] "/Users/soltoffbc/Projects/Computing for Social Sciences/course-site"
```

#### Build a file path

```r
here("static", "extras", "awesome.txt")
## [1] "/Users/soltoffbc/Projects/Computing for Social Sciences/course-site/static/extras/awesome.txt"
cat(readLines(here("static", "extras", "awesome.txt")))
## OMG this is so awesome!
```
    
--

#### What if we change the working directory?

```r
setwd(here("static"))
getwd()
## [1] "/Users/soltoffbc/Projects/Computing for Social Sciences/course-site/static"
cat(readLines(here("static", "extras", "awesome.txt")))
## OMG this is so awesome!
```

---

## Filepaths and Quarto documents

```
data/
  scotus.csv
analysis/
  exploratory-analysis.qmd
final-report.qmd
scotus.Rproj
```

* `.qmd` and assumption of working directory
* Run `read_csv("data/scotus.csv")`
* Run `read_csv(here("data", "scotus.csv"))`

---

## Here's a GIF of Nicolas Cage

---

# Personal R admin

---

## R startup procedures

* Customized startup
* `.Renviron`
* `.Rprofile`

---

## `.Renviron`

* Define sensitive information
* Set R specific environmental variables
* Does not use R code syntax
* `usethis::edit_r_environ()`

## Example `.Renviron`

```shell
R_HISTSIZE=100000
GITHUB_PAT=abc123
R_LIBS_USER=~/R/%p/%v
```

---

## `.Rprofile`

* R code to run when R starts up
* Runs after `.Renviron`
* Multiple `.Rprofile` files
    * Home directory (`~/.Rprofile`)
    * Each R Project folder
* `usethis::edit_r_profile()`

---

## Common items in `.Rprofile`

1. Set a default CRAN mirror
1. Write a welcome message
1. Customize their R prompt
1. Change options, screen width, numeric display
1. Store API keys/tokens that are necessary for only a single project

---

## Git tracking of `.Rprofile`

---

## A couple of things America got right: [cars and freedom](https://www.youtube.com/watch?v=OnQXRxW9VcQ)