final prerelease check

[donyunardi]

• Run R CMD check --as-cran make sure everything pass
• Run urlchecker::url_check() to identify broken links and fix
• Make sure all teal.* mentions are lower-cased and quoted
• Make Sure inst/WORDLIST is minimalized
• Make sure each link to our documentation hosted with pkgdown on github pages do not have /main/ in the address but it has /latest/ instead, so we always expose the documentation of the latest release and not what's currently on main branch but not yet released
• You have checked the Package Release Template https://github.com/insightsengineering/teal.reporter/pull/205/files

[donyunardi]

I got this NOTE:

I noticed teal.code and teal.data are in Imports.

teal.modules.clinical/DESCRIPTION

Lines 55 to 56 in 8469543

	teal.code (>= 0.5.0),
	teal.data (>= 0.4.0),

The current states are:

• tmc Depends on teal:
  

  teal.modules.clinical/DESCRIPTION

  Line 31 in 8469543

  teal (>= 0.15.0),

• teal Depends on teal.data:
  https://github.com/insightsengineering/teal/blob/edfa9b459602563b99b3be3e7275fa9393e2a989/DESCRIPTION#L39

• and teal.data Depends on teal.code:
  https://github.com/insightsengineering/teal.data/blob/0b154de9a4b3d3ca834d8381b41edc2aecdbc508/DESCRIPTION#L32

So, we should be able to move teal.code and teal.data to Suggest for tmc.
I tested this and it looks like this now:

@Melkiades @m7pr @shajoezhu thoughts? Or am I missing something?

[donyunardi]

Also, does anyone know if having too many imports would be a blocker for CRAN?

[chlebowa]

teal.code::eval_code is used throughout all modules. teal.data::join_keys is used a few times. The imports are valid.

[shajoezhu]

i think we can drop formatters and rtables from the list too, since they depends on tern, and comes with dependencies.

[shajoezhu]

this is a working document

broom (>= 0.7.10),
checkmate (>= 2.1.0),
dplyr (>= 1.0.5),
DT (>= 0.13),

drop formatters (>= 0.5.5),
maybe drop? ggplot2 (>= 3.4.0),
ggrepel,
grid,
lifecycle (>= 0.2.0),
logger (>= 0.2.0),
methods,
rlistings (>= 0.2.7),
rmarkdown (>= 2.19),
drop rtables (>= 0.6.6),
scales,
dependencie from teal? can be dropped? shiny (>= 1.6.0),
shinyjs,
shinyvalidate,
shinyWidgets (>= 0.5.1),
stats,
drop teal.code (>= 0.5.0),
drop teal.data (>= 0.4.0),
teal.logger (>= 0.1.1),
teal.reporter (>= 0.2.1),
teal.widgets (>= 0.4.0),
tern.gee (>= 0.1.3),
tern.mmrm (>= 0.3.0),
utils,
vistime

[chlebowa]

i think we can drop formatters and rtables from the list too, since they depends on tern, and comes with dependencies.

Actually, onlyvar_labels is used from formatters. We have copied that function to teal.data as col_labels. Perhaps the latter should be used here?

[m7pr]

teal.code::eval_code is used throughout all modules. teal.data::join_keys is used a few times. The imports are valid.

Yes, those were valid Imports.

Also, does anyone know if having too many imports would be a blocker for CRAN?

I don't think this is a blocker.

[m7pr]

I actually wonder about rtables. since tern is in Depends, and rtables are in Depends of tern. so when you load teal.modules.clinical rtables is loaded because of tern being loaded.

[m7pr]

I'm actually having second thoughts on teal.data and teal.code. The reasoning behind Dony logic is valid. I think you can still use teal.code::eval_code even when teal.code is only in Suggests, because it will be loaded by teal that loads teal.data that loads teal.code. And Dony showed that tests pass when you move teal.data and teal.code out of Imports.

[chlebowa]

I think you mean installed, not loaded. Dependency specification in DESCRIPTION is about installing first and foremost (attaching Depends dependencies comes extra) and as such all dependencies that are directly used should be listed.

I also don't think it's a blocker and if it is we can do selective imports a la @importFrom teal.code eval_code, though I would rather avoid it.