ggtedious

Welcome to the ggtedious workshop!

Let’s build a cool ggplot2 extension function. And then let’s put it in a package. And test! Maybe it will be tedious, but more fun in good company?

‘Testing your code can be painful and tedious, but it greatly increases the quality of your code.’ - testthat introduction (probably Hadley Wickham)

In this workshop, we’ll build and strengthen package building and test writing muscles.

https://angeladuckworth.com/grit-scale/

Meeting objectives:

  1. Practice a compute_group easy geom extension by creating geom_post().
  2. Put them in a package using best practices.
  3. Meet like-minded stats educators, ggplot2 extenders, and package developers.

Prerequisite:

Having written a ‘compute group’ geom extension. See: https://evamaerey.github.io/mytidytuesday/2022-01-03-easy-geom-recipes/easy_geom_recipes.html Seasoned R/ggplot2 users mostly spent ~ 15 minutes on each recipe.

Part 1. Work on functionality

Step 0. Do it with base ggplot2 can become problem statement later

prize_wheel <- data.frame(probs = c(.7, .2, .1), payout = c(0, 1, 5))

library(ggplot2)
ggplot(prize_wheel) + 
  aes(x = payout, y = probs) + 
  geom_point() + 
  aes(xend = payout, yend = 0) + 
  geom_segment()

Step 0.b Write like-to-have code code chunk option eval = F

# would show just line
ggplot(prize_wheel) + 
  aes(x = payout, y = probs) + 
  geom_post()

# line and dot
ggplot(prize_wheel) + 
  aes(x = payout, y = probs) + 
  geom_point() + 
  geom_lollipop()

Step 1. Write compute group function and test

reference: https://evamaerey.github.io/mytidytuesday/2022-01-03-easy-geom-recipes/easy_geom_recipes.html

Step 2. Pass to ggproto object

Step 3. Write geom_post() functions

Step 3.1 Bonus, write geom_lollipop with list mechanism

geom_lollipop <- function(...){

  list(  
  geom_post(...),
  geom_point(...)
  )
  
}

Step 4. Test it out enjoy! (possibly basis of examples and tests)

Step 5. Write messages/warnings etc in the function

Part II. Packaging and documentation 🚧 ✅

Phase 1. Minimal working package

Bit A. Created package archetecture, running devtools::create(".") in interactive session. 🚧 ✅

devtools::create(".")

Bit B. Added roxygen skeleton? 🚧 ✅

Use a roxygen skeleton for auto documentation and making sure proposed functions are exported. Generally, early on, I don’t do much (anything) in terms of filling in the skeleton for documentation, because things may change.

Bit C. Managed dependencies ? 🚧 ✅

Package dependencies managed, i.e. depend::function() in proposed functions and declared in the DESCRIPTION

usethis::use_package("ggplot2")

Bit D. Moved functions R folder? 🚧 ✅

Use new {readme2pkg} function to do this from readme…

library(tidyverse)
readme2pkg::chunk_to_r("geom_post")

Bit E. Run devtools::check() and addressed errors. 🚧 ✅

devtools::check(pkg = ".")

Bit F. Build package 🚧 ✅

devtools::build()

Bit G. Write traditional README that uses built package (also serves as a test of build. 🚧 ✅

The goal of the {ggtedious} package is to make it easy to draw posts (and to learn about package building and testing)

Install package with:

remotes::install_github("EvaMaeRey/ggtedious")

Once functions are exported you can remove go to two colons, and when things are are really finalized, then go without colons (and rearrange your readme…)

library(ggtedious)

Bit H. Chosen a license? 🚧 ✅

usethis::use_mit_license()

Bit I. Add lifecycle badge (experimental) 🚧 ✅

usethis::use_lifecycle_badge("experimental")

Phase 2: Listen & iterate 🚧 ✅

Try to get feedback from experts on API, implementation, default decisions. Is there already work that solves this problem?

Phase 3: Let things settle

Bit A. Settle on examples. Put them in the roxygen skeleton and readme. 🚧 ✅

Bit B. Written formal tests of functions and save to test that folders 🚧 ✅

That would look like this…

library(testthat)

test_that("calc times 2 works", {
  expect_equal(times_two(4), 8)
  expect_equal(times_two(5), 10)
  
})

readme2pkg::chunk_to_tests_testthat("test_calc_times_two_works")

Bit C. Added a description and author information in the DESCRIPTION file 🚧 ✅

Bit D. Addressed all notes, warnings and errors. 🚧

Phase 4. Promote to wider audience…

Bit A. Package website built? ✅

usethis::use_pkgdown()
pkgdown::build_site()

Bit B. Package website deployed? 🚧 ✅

Phase 5: Harden/commit

Submit to CRAN/RUniverse? 🚧

Appendix: Reports, Environment

Edited Description file?

readLines("DESCRIPTION")