Closed salim-b closed 3 years ago
Do we know why tests are failing?
No clue, they only fail on obsolete R versions (< 3.6) with the following error:
Error in gsub("\r\n", "\n", str, fixed = TRUE) :
input string 1 is invalid in this locale
Calls: <Anonymous> -> new_rcmdcheck -> win2unix -> gsub
So the error occurs in rcmdcheck:::win2unix()
. This loosely related SO answer seems to indicate it might have to do with some strange locale issue.
Thanks for looking into this. FWIW, I pushed a minor commit rolling back your change to DESCRIPTION, and rebuilding the docs with the release verision of Roxygen, and the results were the same.
I'm annoyed by this, because I like that we support old versions (not "obsolete"!). But my guess is that this is probably not a big problem and that this is more to do with the testing tooling chain than with the actual functionality of the package.
Still need to investigate.
Good news! The tests on the repo seem to have magically sorted themselves out. Why? Who knows!
But I filed a bug report on the rcmdcheck
repository and the maintainer thinks this is actually a bug on their end. So think we can stop worrying about this for now.
Thanks!
Do we know why tests are failing?