template for CRAN release

[abhsarma]

(this issue is never closed, just re-used for the next release)

• check for existing problems on CRAN checks of multiverse
• double-check CRAN policies
• check package for large files: output from Sys.setenv(NOT_CRAN = "true"); devtools::build() is less than 5MB
• remove any Remotes in DESCRIPTION and downgrade those packages back to CRAN versions
• update R and all packages
  • devtools::install_dev_deps()
  • update.packages(ask = FALSE)
• bump version and date
  • in DESCRIPTION
  • in NEWS
• devtools::check(remote = TRUE, manual = TRUE, env_vars = c(NOT_CRAN = "false"))
• run reverse dependency check
• update NEWS
• spelling::spell_check_package()
• spelling::spell_check_package(use_wordlist = FALSE) then look at title and description (if those have changed lately)
• devtools::document()
• devtools::install()
• rebuild README (knit)
• pkgdown::clean_site()
• Sys.setenv(NOT_CRAN = "true"); pkgdown::build_site(run_dont_run = TRUE)
• Sys.setenv(NOT_CRAN = "true"); devtools::check(remote = TRUE, manual = TRUE, run_dont_test = TRUE)
• remove font?kit files from docs
• Sys.setenv(NOT_CRAN = "false"); devtools::check(remote = TRUE, manual = TRUE, env_vars = c(NOT_CRAN = "false"))
  • R-release, on Windows
  • Whatever platforms on Github actions
• Sys.setenv(NOT_CRAN = "true"); devtools::check_win_release()
• Sys.setenv(NOT_CRAN = "true"); devtools::check_win_devel()
• Sys.setenv(NOT_CRAN = "true"); devtools::check_win_oldrelease()
• write cran-comments.md
• Sys.setenv(NOT_CRAN = "true"); devtools::release()

After package goes live on CRAN:

• merge dev onto master and cran
• tag release
• double check zenodo entry
• authors
• license (GPL >= 3)
• bump to a new dev version on dev branch

Template loosely based on mjskay/ggdist#12

[abhsarma]

devtools::check(remote = TRUE, manual = TRUE)

• checking CRAN incoming feasibility ... NOTE:
  Maintainer: ‘Abhraneel Sarma abhraneel@u.northwestern.edu’
  New submission

rhub::check(platform = "solaris-x86-patched", env_vars = c("_R_CHECK_FORCE_SUGGESTS_" = "false"))

• checking re-building of vignette outputs ...Build timed out (after 10 minutes). Marking the build as failed.
  Build was aborted
  https://builder.r-hub.io/status/multiverse_0.4.0.tar.gz-44f9f29f5eac4605a7895f239d590dc2

[abhsarma]

@mjskay not sure what to do with the rhub issue in the previous comment --- thoughts?

[mjskay]

how long does it take to build your vignettes? you might want to prevent the largest vignettes from building their output when on cran. this is what I do with the tidybayes vignettes --- they are configured not to evaluate their code chunks on cran. This also solves the VignetteBuilder note issue.

You can see the parameters used in tidybayes to do this here:
https://github.com/mjskay/tidybayes/blob/841b23b6c2cf75c300cb816a939808f21c19b75f/vignettes/tidybayes.Rmd#L9-L10

and here:
https://github.com/mjskay/tidybayes/blob/841b23b6c2cf75c300cb816a939808f21c19b75f/vignettes/tidybayes.Rmd#L41

[mjskay]

(this does complicate building and testing for CRAN, see the stuff I have to do in the tidybayes issue turning on and off the NOT_CRAN environment variable: mjskay/tidybayes#112)

[abhsarma]

Don't the vignettes need to be built for CRAN? or do those get uploaded directly and not generated by CRAN directly?

[mjskay]

there are two points where vignettes are built: once when you build them for upload, and once during the CHECK on CRAN so that it can ensure that the code in vignettes runs. The approach I use with tidybayes makes it so that when you build the vignettes locally for upload the code chunks are run (so the vignettes users will see will have code output and plots in them), but disables code chunk execution on CRAN (so CRAN can "build" the vignettes but most of the code is skipped so it doesn't take very long). This ensures the checks don't time out on CRAN but the vignettes still have all their output (e.g. this tidybayes vignette on CRAN has all its output even though I disabled building on that vignette on CRAN)

[mjskay]

This does lead to multiple steps in my CRAN release TODO where I check the package either as CRAN or not as CRAN since the outputs can be different (and sometimes one will error while the other doesn't). Hence all the manipulation of the NOT_CRAN environment variable in this issue.

[mjskay]

basically, having long-running vignettes is a pain in the ass, and I'm not sure if there's a better solution than what I've been doing. If you find one I'd love to hear it