Release glue 1.8.0

[jennybc]

Prepare for release:

• git pull
• Check current CRAN check results
• Polish NEWS
• urlchecker::url_check()
• devtools::build_readme()
• devtools::check(remote = TRUE, manual = TRUE)
• devtools::check_win_devel()
• revdepcheck::cloud_check()
• Update cran-comments.md
• git push

Submit to CRAN:

• usethis::use_version('minor')
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• Accepted 🎉
• usethis::use_github_release()
• usethis::use_dev_version(push = TRUE)

[jennybc]

I'm going to come back and do some analysis and commentary here. Don't expect these links to work, this is copy/pasted from revdeps/README.md. The links should work from there.

Revdeps

Failed to check (3)

package	version	error	warning	note
multinma	0.7.2	1
params	0.7.3	+1
TriDimRegression	1.0.2	1

New problems (11)

package	version	error	warning	note
adoptr	1.1.0	+3
atrrr	0.0.3	+1
codebook	0.9.2	+1		3
gtreg	0.4.0	+2
gtsummary	2.0.2	+2
logger	0.3.0	+1		1
odin	1.2.5	+3
roxytypes	0.1.0	+2		1
shinyCohortBuilder	0.2.1	+2		1
sqltargets	0.2.0	+1
table.glue	0.0.3	+3		1

[jennybc]

Notified based on recent revdep findings:

• adoptr PR 2024-09-19 Prefer glue_data() when data might be list-ish, not an actual environment optad/adoptr#12
• atrrr PR 2024-09-19 Prefer glue_data() when data might be list-ish, not an actual environment JBGruber/atrrr#18
• gtreg I suspect this is an indirect, false-ish positive, coming via gtsummary. Tagged maintainer in the gtsummary issue.
• gtsummary Issue opened 2024-09-19 Failure with dev glue ddsjoberg/gtsummary#1989
• logger Issue opened 2024-09-19 Failure with dev glue daroczig/logger#214
• roxytypes PR 2024-09-19 Prefer glue_data() when data might be list-ish, not an actual environment openpharma/roxytypes#15
• sqltargets PR 2024-09-19 Prefer glue_data_sql() when data is list-ish, not an actual environment daranzolin/sqltargets#22

Notice given Dec 2023 (check means it's been updated on CRAN):

• codebook Issue opened in December 2023. No response. Failure with dev glue rubenarslan/codebook#77
• odin: PR made December 2023 and merged in January 2024, but not yet released on CRAN. 2nd heads up given. Prefer glue_data() when data might be list-ish, not an actual environment mrc-ide/odin#310
• params: PR made in December 2023. No response. Prefer glue_data() when data might be list-ish, not an actual environment sahilseth/params#15
• shinyCohortBuilder: Not on GitHub. Email sent maintainer in December 2023. No response.
• table.glue PR made and merged December 2023, but not yet released on CRAN. 2nd heads up given. Prefer glue_data() when the data source is a list bcjaeger/table.glue#1

---

multinma: Failed to install in the cloud checks, with both CRAN and dev glue. I also can't find any use of the off-label use of glue(.envir =). I think this is a false positive.

I was able to install it locally, but there were many, many compilation warnings and it is extremely slow to build. Likewise R CMD check took a very long time. Test results: [ FAIL 12 | WARN 15 | SKIP 1 | PASS 2228 ]. Tests produce lots of warnings about "divergent transitions after warmup".

Test failures (don't look related to glue):

══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-example_parkinsons.R:531:3'): RE relative effects ────────────
arm_releff_RE$`97.5%` not equivalent to tsd_RE$upper.
1/4 mismatches
[4] 0.64 - 0.69 == -0.0503
── Failure ('test-example_parkinsons.R:543:3'): RE heterogeneity SD ────────────
arm_tau$`97.5%` not equivalent to 1.55.
1/1 mismatches
[1] 1.41 - 1.55 == -0.141
── Failure ('test-example_parkinsons.R:628:3'): RE relative effects ────────────
contr_releff_RE$`97.5%` not equivalent to tsd_RE$upper.
1/4 mismatches
[4] 0.606 - 0.69 == -0.0835
── Failure ('test-example_parkinsons.R:661:3'): RE predicted probabilities at 3 years ──
contr_pred_RE$`97.5%` not equivalent to tsd_pred_RE$upper.
2/5 mismatches (average diff: 0.0667)
[1] -3.19e-05 - 0.05 == -0.0500
[5] -6.34e-02 - 0.02 == -0.0834
── Failure ('test-example_parkinsons.R:747:3'): Analysis of reordered contrast data gives same results ──
predict(...) not equal to `contr_pred_RE`.
Component "97.5%": Mean relative difference: 0.08013476
── Failure ('test-example_parkinsons.R:754:3'): Analysis of reordered contrast data gives same results ──
summary(reorder_fit_RE, pars = "tau") not equal to `contr_tau`.
Component "sd": Mean relative difference: 0.07915171
Component "97.5%": Mean relative difference: 0.05379682
── Failure ('test-example_parkinsons.R:780:3'): RE relative effects ────────────
mix_releff_RE$sd not equivalent to tsd_RE$sd.
1/4 mismatches
[4] 0.726 - 0.8 == -0.0743
── Failure ('test-example_parkinsons.R:792:3'): RE heterogeneity SD ────────────
mix_tau$sd not equivalent to 0.43.
1/1 mismatches
[1] 0.368 - 0.43 == -0.0621
── Failure ('test-example_parkinsons.R:794:3'): RE heterogeneity SD ────────────
mix_tau$`97.5%` not equivalent to 1.55.
1/1 mismatches
[1] 1.37 - 1.55 == -0.181
── Failure ('test-example_parkinsons.R:813:3'): RE predicted probabilities at 3 years ──
mix_pred_RE$sd not equivalent to tsd_pred_RE$sd.
1/5 mismatches
[5] 0.759 - 0.83 == -0.0715
── Failure ('test-example_parkinsons.R:815:3'): RE predicted probabilities at 3 years ──
mix_pred_RE$`97.5%` not equivalent to tsd_pred_RE$upper.
1/5 mismatches
[5] -0.0731 - 0.02 == -0.0931
── Failure ('test-example_transfusion.R:169:3'): Informative RE heterogeneity variance ──
tr_RE_inf_var$`97.5%` not equivalent to 1.84.
1/1 mismatches
[1] 1.74 - 1.84 == -0.104

---

TriDimRegression: Failed to install in the cloud checks, with both CRAN and dev glue, which was apparently also the case with the previous glue release + revdep checks. I also can't find any use of the off-label use of glue(.envir =). I think this is a false positive.

Installation failure from cloud check looks like this:

 installing *source* package ‘TriDimRegression’ ...
** package ‘TriDimRegression’ successfully unpacked and MD5 sums checked
** using staged installation
Error in loadNamespace(x) : there is no package called ‘rstantools’
Calls: loadNamespace -> withRestarts -> withOneRestart -> doWithOneRestart
Execution halted
ERROR: configuration failed for package ‘TriDimRegression’
* removing ‘/tmp/workdir/TriDimRegression/new/TriDimRegression.Rcheck/TriDimRegression’

[jennybc]

Revdep failures found by CRAN (mostly expected, but I'll need to do a bit of research on a couple that are new to me):

*** Changes to worse in reverse dependencies ***
Debian: <https://win-builder.r-project.org/incoming_pretest/glue_1.8.0_20240927_180139/reverseDependencies/summary.txt>
adoptr, atrrr, attrib, codebook, gtreg, gtsummary, logger, params, shinyCohortBuilder, smdi, sqltargets, table.glue

[jennybc]

I expected these. Check mark means the log also looks as I expect:

• adoptr: CRAN's results also include some vignette problems I didn't see, but they seem to originate from the same problem.
• atrrr
• codebook: In addition to the expected .envir thing, both CRAN and I see other failures, such as this from rebuilding vignettes: 'there is no package called 'GGally''.
• gtreg
• gtsummary
• logger
• params
• shinyCohortBuilder
• sqltargets
• table.glue: Hmmm I thought this one had been fixed and released.

New to me:

• attrib: Looks like a false positive. A brittle test, something to do with a numeric expectation.
• smdi: See below. An indirect failure coming via gtsummary.

[ddsjoberg]

If it helps, smdi depends on gtsummary (Suggests not Imports), and could be the reason for the failure.