moodymudskipper / dubioustests Goto Github PK

It is better used defined as

?~ <- function(expr) eval.parent(substitute(rlang::as_function(~expr)))

Moreover we'll get better looking aligned code, and bonus highlighting, by using ?T~, and then we can easily define ?F~ and make expect_false() explicit rather than use ?~ ! expr() which is more confusing.

We should think about general conventions too.

?T~ x == y
?T: x == y
T? x == y

I think the last one is really nice, but we need to discard ?fun? operators as it will make things confusing.

Then we'll have something like this :

"this works" %?% {
  T? is.numeric(x) # expect_true
  F? is.numeric(x) # expect_false
  E? x ? y         # expect_equal, very similar to `T? all.equal(x,y)`
  I? x ? y         # expect_identical, very similar to `I? all.equal(x,y)`
  # every `testthat::expect_<foo>()` function has a counterpart, even those who have a one character shortcut
  length? x ? 3
  warning? x ? "foo"
  message? x ? "foo"
  #...
}