additional conditionals in tests #24
Conversation
| @@ -22,6 +24,8 @@ expect_error(quadfn(5, 1:5), "NULL value passed as symbol address") | |||
| gc() | |||
| expect_false(file.exists(environment(quadfn)$libLFile)) | |||
|
|
|||
| if (isM1) exit_file("Skip remainer") | |||
There was a problem hiding this comment.
Is there a function to skip a single test in tinytest? The CRAN check said that one test out of 30 failed, so we could keep the others.
There was a problem hiding this comment.
from
https://www.stats.ox.ac.uk/pub/bdr/M1mac/inline.out
Running test_utilities.R.............. 13 tests �[0;31m1 fails�[0m �[0;34m2.8s�[0m
----- FAILED[xcpt]: test_utilities.R<35--38>
call| expect_error(current = moveDLL(quadfn, name = "testname", directory = tempdir(),
call| unload = TRUE, overwrite = TRUE), pattern = "file can not be copied both 'from' and 'to'")
diff| No error
Error: 1 out of 30 tests failed
Execution halted
* checking PDF version of manual ... OK
|
But we cannot test on his M1mac so I suggest to not play games. |
|
Actually, the more I think about it the less do I think that it is helpful to do an upload to CRAN just to skip the failing tests. Yes, it would likely give us a clean check page, but in case someone would like to actually use that functionality on affected systems, the warning signs would be gone... But it's your call, you are corresponding with the CRAN team. |
|
Yes, I plan to upload with these changes as I do not intend to have the package tossed off CRAN. |
|
OK then, fingers crossed that nothing new and unforeseen will surface! |
This PR attempts to address the failures on Solaris and M1mac.
On Solaris, we segfault it one file. This can be replicated at RHub using their Solaris machine. We detect the test file and skip right before this occurred. The modified package passes at RHub.
On M1mac, we also die in the last of the three test files. We now skip tests in that file too right before it occurs. We cannot test this.
@jranke : comments / review appreciated