breakerofchains

Break your chain at the cursor line. Run the first bit. See the output. Be free.

Installation

# install.packages("remotes")
remotes::install_github("MilesMcBain/breakerofchains")

Usage

Say you had:

library(tidyverse)

star_plot <-
  starwars %>%
  group_by(species, sex) %>%
  select(height, mass) %>%
  summarise(
    height = mean(height, na.rm = TRUE),
    mass = mean(mass, na.rm = TRUE)
  ) %>%
  ggplot(aes(x = height, y = mass)) +
  geom_point()

1. Pop your cursor on line you want to run up to. e.g. select(height, mass).

2. Invoke the RStudio Addin Break chain and run to cursor

3. Code is run from start of chain up to your cursor line, and result is printed in the console:

starwars %>%
  group_by(species, sex) %>%
  select(height, mass)
#> Adding missing grouping variables: `species`, `sex`
#> # A tibble: 87 × 4
#> # Groups:   species, sex [41]
#>    species sex    height  mass
#>    <chr>   <chr>   <int> <dbl>
#>  1 Human   male      172    77
#>  2 Droid   none      167    75
#>  3 Droid   none       96    32
#>  4 Human   male      202   136
#>  5 Human   female    150    49
#>  6 Human   male      178   120
#>  7 Human   female    165    75
#>  8 Droid   none       97    32
#>  9 Human   male      183    84
#> 10 Human   male      182    77
#> # … with 77 more rows

with a stored result available in .chain:

glimpse(.chain)
#> Rows: 87
#> Columns: 4
#> Groups: species, sex [41]
#> $ species <chr> "Human", "Droid", "Droid", "Human", "Human", "Human", "Human",…
#> $ sex     <chr> "male", "none", "none", "male", "female", "male", "female", "n…
#> $ height  <int> 172, 167, 96, 202, 150, 178, 165, 97, 183, 182, 188, 180, 228,…
#> $ mass    <dbl> 77.0, 75.0, 32.0, 136.0, 49.0, 120.0, 75.0, 32.0, 84.0, 77.0, …

For pipe chains Base pipe |> is supported, but chains can also be broken at lines ending in any %% infix, and any math/logic infix. So you can break ggplot2 layers chained with + this way too.

Stored result .chain

By default the result of the last broken chain is saved in your environment as the variable .chain so you can immediately start passing it to further diagnostics. I’ve found this is nicer than .Last.value which is easy to accidentally overwrite, and has a hard to remember the capitalisation scheme.

Disable this behaviour with options(breakerofchains_store_result = FALSE)

Keybindings

• RStudio: addins can be bound to keys using the keybinding menu.
• VSCode: create a binding for your keybindings.json like:

[
    {
        "description": "run breakerofchains",
        "key": "ctrl+shift+b",
        "command": "r.runCommand",
        "when": "editorTextFocus",
        "args": "breakerofchains::break_chain()"
    },
]

Pitfalls

Since R’s parser is used to help figure out where the chain starts, the process will fail if any of the code above the cursor is invalid - even code not in the chain.

For Rmd documents only code in the current chunk is parsed.

Extending it yourself

break_chain() returns the result of the chain evaluation invisibly, so you can build your own shortcuts that do something with the result other than print it to the console. E.g. View(break_chain()) See break_chain and NEWS.md for more info.