`dwplot` not working when `style = "distribution"`

[IndrajeetPatil]

library(dotwhisker)
#> Loading required package: ggplot2

# run a regression model
mod <- lm(mpg ~ wt + cyl + disp + gear, data = mtcars)

# draw a dot-and-whisker plot
dwplot(mod, style = "distribution")
#> Error: Problem with `mutate()` input `n`.
#> x could not find function "n"
#> ℹ Input `n` is `1:n()`.
#> ℹ The error occurred in group 1: term = "cyl", model = "one".

^{Created on 2021-04-14 by the reprex package (v2.0.0)}

Trace:

> rlang::last_trace()
<error/dplyr:::mutate_error>
Problem with `mutate()` input `n`.
x could not find function "n"
ℹ Input `n` is `1:n()`.
ℹ The error occurred in group 1: term = "cyl", model = "one".
Backtrace:
     █
  1. ├─dotwhisker::dwplot(mod, style = "distribution")
  2. │ └─`%>%`(...)
  3. ├─dplyr::filter(., !is.na(estimate))
  4. ├─dplyr::mutate(...)
  5. ├─dplyr:::mutate.data.frame(...)
  6. │ └─dplyr:::mutate_cols(.data, ...)
  7. │   ├─base::withCallingHandlers(...)
  8. │   └─mask$eval_all_mutate(quo)
  9. └─base::.handleSimpleError(...)
 10.   └─dplyr:::h(simpleError(msg, call))
<error/simpleError>
could not find function "n"

[sammo3182]

Ahm...I can't reproduce the error Indrajee @IndrajeetPatil. I got the following outcome (you'll see I trunked the disp for its coefficient to squeeze the rest to be seen. But plot is still doable. With a wide guess, maybe some other packages overwrote the dplyr::mutate function? Free to reopen the issue if the problem still exists.

library(dotwhisker)

mod <- lm(mpg ~ wt + cyl + gear, data = mtcars)

# draw a dot-and-whisker plot
dwplot(mod, style = "distribution")

[IndrajeetPatil]

Thanks for the response.

I can still reproduce this error:

library(dotwhisker)
#> Loading required package: ggplot2

mod <- lm(mpg ~ wt + cyl + gear, data = mtcars)

# draw a dot-and-whisker plot
dwplot(mod, style = "distribution")
#> Error: Problem with `mutate()` column `n`.
#> ℹ `n = 1:n()`.
#> x could not find function "n"
#> ℹ The error occurred in group 1: term = "cyl", model = one.

^{Created on 2021-07-17 by the reprex package (v2.0.0)}

The way to make it work is by explicitly loading dplyr:

library(dotwhisker)
#> Loading required package: ggplot2
library(dplyr)

mod <- lm(mpg ~ wt + cyl + gear, data = mtcars)

# draw a dot-and-whisker plot
dwplot(mod, style = "distribution")

^{Created on 2021-07-17 by the reprex package (v2.0.0)}

The problem is that you are using dplyr::n() function somewhere but not using either namespace :: or importing it in your package namespace. So the user needs to do it.

[sammo3182]

Ah! Got it! I've updated the namespace in the dev version cc2d53c14b3cb48591059c1ea1a8d4b9203d83c3. I've recorded the bug in the News.md for the next update. Thanks for pointing that out!

[IndrajeetPatil]

One more thing: This produces a cryptic warning:

#> Warning in if (is.na(out)) {: the condition has length > 1 and only the first
#> element will be used

Reproducible example:

library(dotwhisker)
#> Loading required package: ggplot2
# define a regression model
m<- lm(mpg ~ wt * cyl, data = mtcars)

# draw a dot-and-whisker plot
dwplot(m, style = "distribution")
#> Warning in if (is.na(out)) {: the condition has length > 1 and only the first
#> element will be used

^{Created on 2021-09-13 by the reprex package (v2.0.1)}