Testing failing test for message leads to passing test

[JesseAlderliesten]

I tend to wrap tests for messages or warnings around tests like expect_true() to prevent having to write (and execute) the same test twice. For example:

library(tinytest)

be_FALSE <- function(x) {
  message("I'm FALSE")
  FALSE
}

expect_message(expect_true(be_FALSE(x)))
# ----- PASSED      : <-->
#   call| expect_message(expect_true(be_FALSE(x)))

But in interactive use this indicates the tests pass whereas in reality the expect_true() test fails. However, the tests do fail when using tinytest::run_all(). Re-reading the documentation it seems this wrapping tests is undocumented, so I'm wondering if I should expect it to work at all.

So my question is: can something be done to make the test also fail when run interactively, or should I just change my code to split the tests:

library(tinytest)

expect_true(be_FALSE(x))
# I'm FALSE
# ----- FAILED[data]: <-->
#  call| expect_true(be_FALSE(x))
#  diff| Expected TRUE, got FALSE 

expect_message(be_FALSE(x))
# ----- PASSED      : <-->
#   call| expect_message(be_FALSE(x)) 

I'm using tinytest 1.4.3 with R 4.6.0 on Windows 11 x64