system quoting fails with testthat + CMD check

[scottkosty]

Below is a minimal example:

Put the following in tests/testthat.R:

options(warn = 2)
library(testthat)
test_check("anRpackage", report = "summary")

In tests/testthat/test.R

extcmd <- 'R -e \'print("hello")\''
system(extcmd, intern = TRUE)

Then run R CMD check (either with 3.1 or with the current devel version), I get the following error:

* checking tests ...
  Running ‘testthat.R’
 ERROR
Running the tests in ‘tests/testthat.R’ failed.
Last 13 lines of output:
  > library(testthat)
  > 
  > test_check("anRpackage", report = "summary")
  Loading required package: anRpackage
  Error in file(filename, "r", encoding = encoding) : 
    cannot open the connection
  Calls: local ... eval.parent -> eval -> eval -> eval -> eval -> source -> file
  In addition: Warning message:
  In file(filename, "r", encoding = encoding) :
    cannot open file 'startup.Rs': No such file or directory
  Execution halted
  Error: (converted from warning) running command 'R -e 'print("hello")'' had status 1
  Execution halted

If I source test.R or run R CMD BATCH test.R or put test.R directly in the tests folder and then run R CMD check, it runs fine.

The error is similar to #129 and #86.

I'm using Ubuntu 64-bit, both 13.10 and 14.04.

[eddelbuettel]

It's R. I am getting the same issue if I just try some system("R CMD build ...") via a file in tests/ -- no testthat involved (and I got here via Google...)

Ubuntu 14.04, R 3.1.0.

[scottkosty]

Good to know, Dirk. Let's close this issue then.

[eddelbuettel]

Noooo -- I want it fixed too :-/ We need to find a fix that is orthogonal to whether it gets tickled by devtools or testthat or youknowhat.

[scottkosty]

I agree but from your new information this sounds like an R issue, not a testthat issue right? I certainly don't mind keeping this bug open, but I would not want to clutter testthat with a bug that's not really actionable by its author, unless we consider a work around I suppose.

Don't you think it's better to report this to R core and deal with it there?

[hadley]

This is an R issue. In devtools we have (in the context of setting environment variables):

    # When R CMD check runs tests, it sets R_TESTS. When the tests
    # themeselves run R CMD xxxx, as is the case with the tests in
    # devtools, having R_TESTS set causes errors because it confuses
    # the R subprocesses. Unsetting it here avoids those problems.
    "R_TESTS" = "",

[fabian-s]

Just came here to confirm that adding

Sys.setenv("R_TESTS" = "")

to the testfile seems to take care of these issues.

[dtenenba]

Has this been reported to R-core? If so, where? Thanks.