For instance, @inheritParams ellipsis::dots_empty would generate:

These dots are for future extensions and must be empty.

This way we don't end up with multiple variations of the above.

library(ordinal)
#> Warning: package 'ordinal' was built under R version 3.5.1

fit <- clm(rating ~ temp * contact, data = wine)

broom_confint_terms <- function(x, ...) {
  
  # warn on arguments silently being ignored
  ellipsis::check_dots_used()
  ci <- confint(x, ...)
  
  # confint called on models with a single predictor
  # often returns a named vector rather than a matrix :(
  
  if (is.null(dim(ci))) {
    ci <- matrix(ci, nrow = 1)
    rownames(ci) <- names(coef(x))[1]
  }
  
  ci <- as_tibble(ci, rownames = "term")
  names(ci) <- c("term", "conf.low", "conf.high")
  ci
}

conf.level <- 0.99
broom_confint_terms(fit, level = conf.level)
#> Error in eval_bare(exit_handler, env): object 'ellipsis_eval_bare' not found

confint(fit, level = 0.99)
#>                          0.5 %   99.5 %
#> tempwarm             0.5919548 4.248065
#> contactyes          -0.3059129 3.147587
#> tempwarm:contactyes -2.0408370 2.761428

^{Created on 2019-03-10 by the reprex package (v0.2.1)}

Needed for vctrs

Prepare for release:

• devtools::check()
• devtools::check_win_devel()
• rhub::check_for_cran()
• revdepcheck::revdep_check(num_workers = 4)

Submit to CRAN:

• usethis::use_version('minor')
• Update cran-comments.md
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• Accepted 🎉
• usethis::use_github_release()
• usethis::use_dev_version()

https://github.com/hadley/ellipsis/blob/d1bdd84d057172533eca9e447a8fced9f2e7f81b/R/safe.R#L36-L38

I believe it should call the median function instead of mean.

Should the UI be on.exit(check_dots_unused())? I think hiding the on.exit() makes it harder to read. Also harder to experiment with. Might also lead to bugs by writing things like:

check_dots_unused()

on.exit(cleanup())

Prepare for release:

• devtools::check()
• devtools::check_win_devel()
• rhub::check_for_cran()
• revdepcheck::revdep_check(num_workers = 4)
• Polish NEWS

Submit to CRAN:

• usethis::use_version()
• Update cran-comments.md
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• Accepted 🎉
• usethis::use_github_release()
• usethis::use_dev_version()

IMHO this should not cause a warning, as a was matched by the method, just not used in the body.

library(ellipsis)

foo <- function(x, ...) {
  ellipsis::check_dots_used()
  UseMethod("foo")
}

foo.character <- function(x, ..., a) {
  x
}

foo("hi", a = 1)
#> Warning: Some components of ... were not used: a
#> [1] "hi"

Created on 2018-07-09 by the reprex package (v0.2.0).

• https://twitter.com/dinga92/status/1102614185459048448
• tidyverse/dplyr#4229

I would expect the call to bar() to issue a warning here, because the bar generic is passed an argument a without it being used by the method. Also note you get a warning if from the foo check_dots_used() if you do not explicitly force the promise for a.

library(ellipsis)

foo <- function(x, ...) {
  ellipsis::check_dots_used()
  UseMethod("foo")
}

bar <- function(x, ...) {
  ellipsis::check_dots_used()
  UseMethod("bar")
}

foo.character <- function(x, ..., a = 1) {
  force(a)
  bar(x, ..., a = a)
}

bar.character <- function(x, ...) {
  x
}

foo("hi", a = 1)
#> [1] "hi"

Created on 2018-07-09 by the reprex package (v0.2.0).

If ellipsis is used in low level tools, warning immediately might result in an avalanche of output when the tools are used in loops.

Prepare for release:

• devtools::check()
• devtools::check_win_devel()
• rhub::check_for_cran()
• rhub::check(platform = 'ubuntu-rchk')
• rhub::check_with_sanitizers()

Submit to CRAN:

• usethis::use_version('0.1.0')
• devtools::check_win_devel() (again!)
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• usethis::use_github_release()
• usethis::use_dev_version()

The master branch of this repository will soon be renamed to main, as part of a coordinated change across several GitHub organizations (including, but not limited to: tidyverse, r-lib, tidymodels, and sol-eng). We anticipate this will happen by the end of September 2021.

That will be preceded by a release of the usethis package, which will gain some functionality around detecting and adapting to a renamed default branch. There will also be a blog post at the time of this master --> main change.

The purpose of this issue is to:

• Help us firm up the list of targetted repositories
• Make sure all maintainers are aware of what's coming
• Give us an issue to close when the job is done
• Give us a place to put advice for collaborators re: how to adapt

message id: euphoric_snowdog

Hi,

As stated in the Tidyverse design guide, you cannot use check_dots_used if your function uses enquos(...) or match.call() somewhere.

I think it would be a good thing to add something in the documentation of check_dots_used that clarifies this point. You also might want to state whether this is a permanent limitation or a limitation that could be fixed someday.

Thanks for this useful package

f <- function(x, ...) {
  ellipsis::check_dots_used(action = rlang::warn)
}
f(z = 1)
#> Warning: 1 components of `...` were not used.
#> 
#> We detected these problematic arguments:
#> * `z`
#> 
#> Did you misspecify an argument?

Can we make it this?

#> Warning: 1 components of `f(...)` were not used.

Probably also worth a refresh here to use our more modern style:

#> Warning: Some arguments in `f(...)` were not used.
#> x Unused arguments: `z`
#> i Did you misspecify an argument?

as_factor <- function(x, ...) {
  ellipsis::check_dots_used()
  UseMethod("as_factor")
}

as_factor.character <- function(x, ..., y)  {
  force(y)
  x
}
as_factor("x", y = 1)
#> Warning: Some components of ... were not used: y
#> [1] "x"

^{Created on 2019-02-18 by the reprex package (v0.2.1.9000)}

Since check_dots_used() checks that dots were used, I would expect check_dots_unused() to check the opposite, that dots were passed but not used. check_dots_empty() seems like a better name.

Prepare for release:

• devtools::check()
• devtools::check_win_devel()
• rhub::check_for_cran()
• rhub::check(platform = 'ubuntu-rchk')
• rhub::check_with_sanitizers()
• revdepcheck::revdep_check(num_workers = 4)
• Polish NEWS
• Polish pkgdown reference index
• Draft blog post

Submit to CRAN:

• usethis::use_version('minor')
• Update cran-comments.md
• devtools::submit_cran()
• Approve email

Wait for CRAN...

• Accepted 🎉
• usethis::use_github_release()
• usethis::use_dev_version()
• Finish blog post
• Tweet
• Add link to blog post in pkgdown news menu

We use ellipsis in sparklyr and it's also in CRAN, should we graduate this package from experimental?

For the case where you're using ... simply to force the user to fully name details arguments

This is intended as a heads-up before many upcoming errors in CRAN checks.

Starting from R-devel version "2020-05-09 r78394", CRAN checks produce errors with the same message "'where' must be an environment" (for example, see this log). This comes from base::bquote(...), which is used in ellipsis:::exit_handler(action).

Current CRAN version of 'ellipsis' has implementation of exit_handler() using bquote() with where argument being a list. Recent update in R-devel explicitly prohibited where to be anything other than environment (see this commit on R-devel). This seems to be the root cause of errors.

on_exit() implementation in development version is different and is using eval() (changed in this commit), which might have already fixed the issue, but is not yet on CRAN.

cc @hadley, @lionel- , as this seems to be important ('ellipsis' is used in 'tidyr', which has many reverse dependencies).

Due respect, I'm not sure if this is the best place to ask the question.

Scenario:

• Parent function passes ... to one or more downstream functions.
• Downstream function does not allow ..., and has a lot of arguments (83).
• Parent function would like to allow ... for customization of downstream function.
• Parent function would like to pass only appropriate ... arguments to downstream function, to avoid causing an error.

From section 6.2.4 https://adv-r.hadley.nz/functions.html#function-fundamentals it appears that do.call(function, args) could be used for this purpose. I wonder if some type of function would live in this package?

Example scenario:

my_lil_function <- function(x, ...) {
   mean_function(x, ...)
}
mean_function <- function(x, na.rm=FALSE) {
   mean(x, na.rm=na.rm)
}
# success
my_lil_function(c(1, 2.5, 3, NA), na.rm=TRUE)
# error
my_lil_function(c(1, 2.5, 3, NA), na.rm=TRUE, color="red")
# Error in mean_function(x, ...) : unused argument (color = "red")

Rough idea of workaround:

# new function for this package?
call_fn_ellipsis <- function(FUN, ...) {
   FUN_argnames <- names(formals(FUN));
   if ("..." %in% FUN_argnames) {
      # if FUN allows ... then pass all of ...
      FUN(...)
   } else {
      # if FUN does not allow ... then pass only what it accepts
      arglist <- list(...)
      argkeep <- which(names(arglist) %in% FUN_argnames)
      arguse <- arglist[argkeep]
      do.call(FUN, arguse)
   }
}

# new method would involve using do.call()
my_new_lil_function <- function(x, ...) {
   args_to_send <- c(list(x=x),
      list(...));
   do.call(call_fn_ellipsis,
      c(FUN=mean_function, args_to_send))
}
# now success
my_new_lil_function(c(1, 2.5, 3, NA), na.rm=TRUE, color="red")
# [1] 2.17

Main drivers:

• I don't want to add 83 arguments to my_lil_function() in order to avoid using ...
• I want to use ... for customizations to two different functions called by my_lil_function()
• I do not control the downstream function, so I can't add ... to that function's arguments.

Suggestions, feedback, guidance welcomed!
Maybe there is an R package function that already performs this step.
Thank you!

All non-RStudio employees/contractors have only submitted typo fixes — https://github.com/r-lib/ellipsis/graphs/contributors

I have an S4 class defined as part of a Bioconductor package and some of the slots are used to store tibble objects. I've just made the upgrade to R4.0.2 (from R3.6.3) and am now getting an unexpected warning when retrieving these slots. Strangely this isn't always given as a warning upon the initial call, but is returned every time a resultant object is printed. For an MWE:

options(warn = 2)
library(tibble)
setClass("track", slots = c(x="numeric", y="data.frame"))
myTrack <- new("track", x = -4:4, y = tibble(y = 1))
myTrack
df <- slot(myTrack, "y")
df
traceback()
16: doWithOneRestart(return(expr), restart)
15: withOneRestart(expr, restarts[[1L]])
14: withRestarts({
        .Internal(.signalCondition(cond, message, call))
        .Internal(.dfltWarn(message, call))
    }, muffleWarning = function() NULL)
13: warning(cnd)
12: action(message, .subclass = c(.subclass, "rlib_error_dots"), 
        ...)
11: action_dots(action = action, message = "`...` is not empty.", 
        dot_names = names(dots), note = "These dots only exist to allow future extensions and should be empty.", 
        .subclass = "rlib_error_dots_nonempty")
10: check_dots_empty(action = warn)
9: pillar::colonnade(df, has_row_id = if (star) "*" else TRUE, needs_dots = needs_dots)
8: shrink_mat(df, rows, n, star = has_rownames(x))
7: trunc_mat(x, n = n, width = width, n_extra = n_extra)
6: format.tbl(x, ..., n = n, width = width, n_extra = n_extra)
5: format(x, ..., n = n, width = width, n_extra = n_extra)
4: paste0(..., collapse = "\n")
3: cli::cat_line(format(x, ..., n = n, width = width, n_extra = n_extra))
2: print.tbl(x)
1: (function (x, ...) 
   UseMethod("print"))(x)

Interestingly, the package passes R CMD check despite these warnings being produced by most examples.

R version 4.0.2 (2020-06-22)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 18.04.4 LTS

Matrix products: default
BLAS:   /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.7.1
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.7.1

locale:
 [1] LC_CTYPE=en_AU.UTF-8       LC_NUMERIC=C               LC_TIME=en_AU.UTF-8       
 [4] LC_COLLATE=en_AU.UTF-8     LC_MONETARY=en_AU.UTF-8    LC_MESSAGES=en_AU.UTF-8   
 [7] LC_PAPER=en_AU.UTF-8       LC_NAME=C                  LC_ADDRESS=C              
[10] LC_TELEPHONE=C             LC_MEASUREMENT=en_AU.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] tibble_3.0.2

loaded via a namespace (and not attached):
 [1] compiler_4.0.2   assertthat_0.2.1 magrittr_1.5     ellipsis_0.3.1   cli_2.0.2        tools_4.0.2     
 [7] pillar_1.4.5     glue_1.4.1       rstudioapi_0.11  crayon_1.3.4     utf8_1.1.4       fansi_0.4.1     
[13] vctrs_0.3.1      lifecycle_0.2.0  pkgconfig_2.0.3  rlang_0.4.7  

Thanks in advance