exit_file / exit_if_not not working

[LiNk-NY]

Hi Mark, @markvanderloo

I hope you are well.

I am using tinytest for a Bioconductor package: AnVILAz and I am having difficulty using exit_file() and exit_if_not.
They don't seem to be working properly.
I have three methods of exiting the test file using both exit_file() and exit_if_not():

## method 1
AnVILAz:::.exit_if_not_anvilaz()
> dput(AnVILAz:::.exit_if_not_anvilaz)
function () 
{
    if (!.is_anvil_az()) 
        tinytest::exit_file("Not running on AnVIL Azure workspace")
}

## method 2
if (!AnVILAz:::.is_anvil_az())
    tinytest::exit_file("Not running on AnVIL Azure workspace")

## method 3
tinytest::exit_if_not(
    "Not running on AnVIL Azure workspace" = AnVILAz:::.is_anvil_az()
)

Despite these trips, the tests continue to run when running R CMD check and tinytest::test_package("AnVILAz") :

> tinytest::test_package("AnVILAz")
test_az_copy.R................    0 tests    sh: 1: az: not found
sh: 1: az: not found
Error in system2("az", c("account", "get-access-token"), stdout = TRUE) : 
  error in running command
In addition: Warning message:
In system2("az", c("login", "--identity", "--allow-no-subscriptions")) :
  error in running command

For context, the package is meant to run on an environment with the az command line utility installed.

> packageVersion("tinytest")
[1] '1.4.1'

[markvanderloo]

Yes, this is fixed in the development version. I have to ask one or two authors to update their package on CRAN before I can upload because of some false positives there.

[LiNk-NY]

Thank you @markvanderloo for looking into it!

FWIW, I have tested with the development version with the same result:

> tinytest::run_test_file("inst/tinytest/test_az_copy.R")
test_az_copy.R................    0 tests    sh: 1: az: not found
sh: 1: az: not found
Error in system2("az", c("account", "get-access-token"), stdout = TRUE) : 
  error in running command
In addition: Warning message:
In system2("az", c("login", "--identity", "--allow-no-subscriptions")) :
  error in running command
> packageVersion("tinytest")
[1] '1.4.1.1'

[grantmcdermott]

I believe that we just got bit by this over at tinyplot (per Vincent’s ref link above). Any chance of pushing an update to CRAN now that revdeps would have had time to adjust?