Open aronatkins opened 3 years ago
══ Warnings ════════════════════════════════════════════════════════════════════
── Warning (test-local-repositories.R:8:5): init fails when package not found in any repo ──
path[1]="repo-empty": No such file or directory
Backtrace:
1. base::options(repos = c(CRAN = paste0("file:///", normalizePath("repo-empty")))) test-local-repositories.R:8:4
3. base::normalizePath("repo-empty")
This probably happens because R removes empty directories during build. We could probably subvert it by adding a single empty file or something similar? Or by creating that directory on demand rather than expecting it to exist.
I don't have a great idea on the other issues, though. :-/
The other failures had skip_on_travis
but that doesn't work in GHA. They now have skip_on_ci
as of https://github.com/rstudio/packrat/pull/638.
Sadly, we now see:
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure (test-dependencies.R:58:3): dependencies are discovered in inline R code ──
"emo" %in% packrat:::fileDependencies(emojiRmd) is not TRUE
`actual`: FALSE
`expected`: TRUE
── Failure (test-rmarkdown.R:22:3): We can discover dependencies with an evaluate hook ──
`deps` not equal to c("abc", "def", "ghi", "jkl").
target is NULL, current is character
Unlike the others, these failures do not happen in local "check" runs.