Use expect_no_warning()

[IndrajeetPatil]

Instead of expect_warning() + NA combo.

Automated via:

replace_multiline <- function(dir = "tests/") {
  files <- list.files(dir, pattern = "\\.R$", full.names = TRUE, recursive = TRUE)
  
  pattern <- "(?s)expect_warning\\((.*?)\\,\\s*NA\\)"
  replacement <- "expect_no_warning(\\1)"
  
  for (f in files) {
    txt <- xfun::read_utf8(f)
    content <- paste(txt, collapse = "\n")
    new_content <- gsub(pattern, replacement, content, perl = TRUE)
    
    if (!identical(content, new_content)) {
      new_txt <- unlist(strsplit(new_content, "\n", fixed = TRUE))
      xfun::write_utf8(new_txt, f)
      message("Modified: ", f)
    }
  }
  
  invisible(NULL)
}

replace_multiline("tests/")

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 0d4cc51 is merged into main:

• ✔️cache_applying: 165ms -> 165ms [-3.08%, +2.47%]
• ✔️cache_recording: 559ms -> 559ms [-1.25%, +1.23%]
• ✔️without_cache: 1s -> 997ms [-0.86%, +0.05%]

Further explanation regarding interpretation and methodology can be found in the documentation.

[lorenzwalthert]

Did you do it yourself or instruct chatGPT or similar to do this? 😄

[IndrajeetPatil]

I started with a regex that didn't work and ChatGPT took it from there 😈