Suggesting an additional argument to `isAvailable`

[dipterix]

Many people and package maintainers use isAvailable without wrapping it around try-catch, thinking the function never errors out. However, I found myself in a weird situation where isAvailable errors out if the child processes mess up the system environments.

This issue happens when a job (child 1) wipe out the system environment but launches another rstudio job (child of child 1) from within...

tfile <- tempfile()
writeLines(
  r'(
Sys.unsetenv("RSTUDIOAPI_IPC_REQUESTS_FILE")
Sys.unsetenv("RSTUDIOAPI_IPC_RESPONSE_FILE")
Sys.unsetenv("RSTUDIOAPI_IPC_SHARED_SECRET")
Sys.setenv("RSTUDIO_WORKBENCH_JOB" = "1")
print(rstudioapi::isJob())
print(rstudioapi::isAvailable(child_ok = TRUE))
)', con = tfile
)

source(tfile)
#> [1] TRUE
#> Error in callRemote(sys.call(), parent.frame()) : 
#>   internal error: callRemote() called without remote connection

rstudioapi::jobRunScript(tfile)
#> Error in callRemote(sys.call(), parent.frame()) : 
#>   internal error: callRemote() called without remote connection

Would it be OK to add an additional argument to isAvailable that if the function errors out, show a warning instead of erroring out?