Report rcheck envvars

[tanho63]

Closes #172 - prints _R_CHECK_*, NOT_CRAN, and other related rcmdcheck envvars.

R> rcmdcheck::rcmdcheck("tests/testthat/bad1")
── R CMD build ───────────────────────────────────────────────────────────────────────────────
pdflatex not found! Not building PDF manual.
✔  checking for file ‘.../DESCRIPTION’ ...
─  preparing ‘badpackage’:
✔  checking DESCRIPTION meta-information
─  installing the package to build vignettes
✔  creating vignettes (1.4s)
─  checking for LF line-endings in source and make files and shell scripts (351ms)
─  checking for empty or unneeded directories
─  building ‘badpackage_1.0.0.tar.gz’
   
── R CMD check ───────────────────────────────────────────────────────────────────────────────

─ R CMD check env vars set:
• _R_CHECK_PKG_SIZES_THRESHOLD_ = 142

R> rcmdcheck::rcmdcheck("tests/testthat/bad2")
── R CMD build ───────────────────────────────────────────────────────────────────────────────
pdflatex not found! Not building PDF manual.
✔  checking for file ‘.../DESCRIPTION’ ...
─  preparing ‘badpackage’:
✔  checking DESCRIPTION meta-information ...
─  checking for LF line-endings in source and make files and shell scripts
─  checking for empty or unneeded directories
─  building ‘badpackage_1.0.0.tar.gz’
   
── R CMD check ───────────────────────────────────────────────────────────────────────────────

─ No R CMD check env vars set
─  using log directory ‘/tmp/RtmptQxJYO/file111243a95c5f9/badpackage.Rcheck’

[hadley]

I think you could just print _R_CHECK var here — the idea is that devtools prints the env vars that it sets; rcmdcheck prints all the env vars that affect the operation of R CMD check, regardless of how they've been setting.

[tanho63]

I was thinking that it looked a little duplicative when called from devtools::check() and that we could let it all be handled by rcmdcheck? Here's what it looks like when you call devtools::check() with the PR rcmdcheck + release version devtools (note that it reprints the same variables, basically):

R> devtools::check("~/_nflverse/nflreadr")
══ Documenting ══════════════════════════════════════════════════════════════════════════════
ℹ Updating nflreadr documentation
ℹ Loading nflreadr

══ Building ═════════════════════════════════════════════════════════════════════════════════
Setting env vars:
• CFLAGS    : -Wall -pedantic -fdiagnostics-color=always
• CXXFLAGS  : -Wall -pedantic -fdiagnostics-color=always
• CXX11FLAGS: -Wall -pedantic -fdiagnostics-color=always
• CXX14FLAGS: -Wall -pedantic -fdiagnostics-color=always
• CXX17FLAGS: -Wall -pedantic -fdiagnostics-color=always
• CXX20FLAGS: -Wall -pedantic -fdiagnostics-color=always
── R CMD build ──────────────────────────────────────────────────────────────────────────────
✔  checking for file ‘/home/tan/_nflverse/nflreadr/DESCRIPTION’ ...
─  preparing ‘nflreadr’:
✔  checking DESCRIPTION meta-information ...
─  installing the package to build vignettes
✔  creating vignettes (7s)
─  checking for LF line-endings in source and make files and shell scripts (591ms)
─  checking for empty or unneeded directories
─  building ‘nflreadr_1.5.0.tar.gz’
   
══ Checking ═════════════════════════════════════════════════════════════════════════════════
Setting env vars:
• _R_CHECK_CRAN_INCOMING_USE_ASPELL_           : TRUE
• _R_CHECK_CRAN_INCOMING_REMOTE_               : FALSE
• _R_CHECK_CRAN_INCOMING_                      : FALSE
• _R_CHECK_FORCE_SUGGESTS_                     : FALSE
• _R_CHECK_PACKAGES_USED_IGNORE_UNUSED_IMPORTS_: FALSE
• NOT_CRAN                                     : true
── R CMD check ──────────────────────────────────────────────────────────────────────────────

─ R CMD check env vars set:
• _R_CHECK_CRAN_INCOMING_                       = FALSE
• _R_CHECK_CRAN_INCOMING_REMOTE_                = FALSE
• _R_CHECK_CRAN_INCOMING_USE_ASPELL_            = TRUE
• _R_CHECK_FORCE_SUGGESTS_                      = FALSE
• _R_CHECK_PACKAGES_USED_IGNORE_UNUSED_IMPORTS_ = FALSE
• NOT_CRAN                                      = true

My initial instinct was to make printing the envvars that relate to checking all get printed by rcmdcheck (even if set by devtools or otherwise), so also made a corresponding PR to remove the printing by devtools r-lib/devtools#2621

[tanho63]

as for the RCMDCHECK / R_TESTS grep patterns, I did a search through the rcmdcheck codebase for Sys.getenv https://github.com/search?q=repo%3Ar-lib%2Frcmdcheck%20Sys.getenv&type=code and saw those two patterns being referred to in a handful of places, figured they might as well get printed

[tanho63]

(I'm happy to change to "just" the _R_CHECK ones if we have a stronger preference to have devtools print the vars set by devtools, just explaining the original rationale)

[hadley]

If we had this, we'd drop the R_CHECK_ printing in devtools, but still set the other envvars.

rcmdcheck is used in other places apart from devtools, so the goals are a little different — here I'm worried more about env vars that are set elsewhere (i.e. not by devtools). We can also wait for Gabor to take a look.

[tanho63]

If we had this, we'd drop the R_CHECK_ printing in devtools, but still set the other envvars.

Yeah, I think that's what I tried to implement here (maybe worth eyeballing to confirm?) https://github.com/r-lib/devtools/pull/2621/files