Main

[bbolker]

With the move to GitHub actions I changed skip_on_travis() to skip_on_ci() throughout. I also commented out the scheduled cron test in the workflow (seems like overkill).

I am still having problems passing devtools tests; I think these may be due to sporadic network access problems? Ran devtools::test() twice in a row. First time results were

[ FAIL 4 | WARN 1 | SKIP 1 | PASS 100 ]

second time was

[ FAIL 1 | WARN 1 | SKIP 1 | PASS 110 ]

I don't know if there's a way to make the tests more robust or not (I also noticed that on GitHub actions a different subset of platforms would fail each time ...

[fmichonneau]

the tests produce some false positive because of difference in caching between the data scrapped from the CRAN website and the RDS database that holds the results of the tests. For a few days the number of failing tests was much higher than usual (maybe because of the CRAN vacation?). Things are still not perfect but back to usual lower levels of errors. I want to keep to run the tests regularly on the main repo as it helps me detect when things change on CRAN's side but maybe I could update the GitHub actions file to have the cron jobs not run in forks.