Hadley Wickham
Hadley Wickham
@olivroy IMO the danger there comes from `==`, not `all()`.
I'd love a PR for this (maybe as part of TDD) — for now I think we can go with `required = c("one", "any", "all")`, and I think it's probably...
Coming back to this again, I wonder if it might be better to make these separate functions, like `expect_all_true()` and `expect_any_true()`?
It's worth noting that while we can give a better failure message for the the all variant, we can't for the any variant.
Could you please rework your reproducible example to use the [reprex package](http://reprex.tidyverse.org) ? That makes it easier to see both the input and the output, formatted in such a way...
Somewhat more minimal reprex, but must be run interactively: ```R testthat::expect_message(1, "x", fixed = TRUE) ``` This is going to be a fun one to debug!
Ok, here's a reprex that succinctly illustrates the problem: ``` r testthat::local_edition(2) testthat::expect_message(1, "x", fixed = TRUE) #> Error: 1 did not produce any messages. testthat::local_edition(3) testthat::expect_message(1, "x", fixed =...
Could you have a go at replicating with a simpler data frame, maybe following the advice at https://dbplyr.tidyverse.org/articles/reprex.html? Otherwise, it's unlikely I'll have the time to run this problem down,...
I've closed this issue due to lack of requested reprex. If you still care about this bug, please open a new issue with a reprex.