Fix issue #2624 #322
Fix issue #2624 #322
Conversation
|
Yes, it seems that on Mac it is producing NaN now somewhere instead of NA, but the test is only for some edge case anyway, and I suppose NaN works too. Unfortunately I cannot fix it on Mac. |
|
But looks like I need another review to merge... |
|
It is probably best to add testthat::skip_on_os("mac") to the failing test and then make another test specifically for mac with the NaN while adding testthat::skip_on_os(c("windows", "linux")). Preferably we have all successful tests even if there is a NaN. So then you have one test for windows and linux and one for mac. I'll leave that decision up to you. See jasp-stats/jaspMachineLearning@0ee2d6f for a similar fix. |
| @@ -337,7 +337,7 @@ doeFactorial <- function(jaspResults, dataset, options, ...) { | |||
| df <- .doeRsmCategorical2df(options[["categoricalVariables"]]) | |||
| designSpec <- .doeFactorialGetSelectedDesign(jaspResults, options) | |||
| if (length(designSpec) == 0) { | |||
| return() | |||
| stop("No generator provided.") | |||
There was a problem hiding this comment.
jaspBase:::.quitAnalysis()?
|
I did not know about he skipping. Thank you! I added it. |
fixes https://github.com/jasp-stats/jasp-test-release/issues/2624