Helper functions for PrairieLearn and R as used by STAT430 (during 2018-2020) and now STAT447 along with STAT385.
Here we assume the instructor has created a file called solution.R in the tests directory
containing the reference answer and that the student has submitted a file called submission.R.
See the PrairieLearn docs for more details
on the external autograder.
For a question that test the creation of objects, a test file might look like:
# load packages and extensions
library(tinytest)
using(ttdo)
# source student and reference files
std = plr::source_std(file = "submission.R")
ref = plr::source_ref(file = "solution.R")
## @title Does the vector a have the correct length?
## @score 1
expect_equal_with_diff(length(std$a), length(ref$a))
## @title Does the vector b have the class?
## @score 1
expect_equal_with_diff(dim(std$b), dim(ref$b))For a question that test a function, a test file might look like:
# load packages and extensions
library(tinytest)
using(ttdo)
# source student and reference files
std = plr::source_std(file = "submission.R")
ref = plr::source_ref(file = "solution.R")
## @title Does the function return the correct type with default inputs?
## @score 1
expect_equal_with_diff(typeof(std$fun()), typeof(ref$fun()))
## @title Does the function return the correct object when x = 1?
## @score 1
expect_equal_with_diff(std$fun(x = 1), ref$fun(x = 1))
## @title Does the function return the correct object when x is random?
## @score 1
y = sample(2:10, 1)
expect_equal_with_diff(std$fun(x = y), ref$fun(x = y))These two testing patterns can be combined to evaluate many possible R problems. Additionally, the tests can be grouped arbitrarily (from a single test in each file to every test in one file) to adapt to the author's preferences.
Dirk Eddelbuettel, Alton Barbehenn, and David Dalpiaz
Jeroen Ooms for the excellent unix package, and of course the whole PrairieLearn team for their system