R CMD check failure

[Andrew-Acker125]

We've been running R CMD check on packages for our package registry manager PRISM (Home | PRISM) and noticed that cmdstanr 0.9.0 is failing during tests with 1 failure. This appears related to data.table 1.18.0, as the test passes with an older version of data.table.
The full error output:

checking tests Running testthat.R
Running the tests in tests/testthat.R failed.
Last 13 lines of output:
 3.  testthat (local) .capture(...)
 4.   base::withCallingHandlers(...)
 5.  rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
 6. cmdstanr::read_cmdstan_csv(file_path)
 7. cmdstanr:::read_csv_metadata(output_file)
 8. base::suppressWarnings(...)
 9.  base::withCallingHandlers(...)
 10. data.table::fread(...)
 11. data.table:::stopf(...)
 12. data.table:::raise_condition(...)
[ FAIL 1 | WARN 1 | SKIP 4 | PASS 1918 ]
Error: ! Test failures. Execution halted

Here is the failing test:

test_that("read_cmdstan_csv() fails with empty csv file", {
  file_path <- test_path("resources", "csv", "empty.csv")
  file.create(file_path)
  expect_error(read_cmdstan_csv(file_path),
               "Supplied CSV file is corrupt!")
  file.remove(file_path)
})

Happy to provide any more details or clarity if needed.

[jgabry]

Thanks for reporting this. I'm pretty sure this is the same issue that we fixed in #1127, but we just haven't tagged a new release yet since that was merged. I'm not quite sure when the next release will be, but it's on my to-do list!

[Andrew-Acker125]

Appreciate it @jgabry !

[jgabry]

I'm going to close the issue since the underlying problem was fixed in #1127 (I took a closer look an confirmed), but I'll be working on getting cmdstanr v1.0 out, which will include this fix. The timing just depends on how quickly I can get to a few of the remaining issues we want to close before that release, but we're mostly there. Also, PRISM is very cool!