`geom_beeswarm()` gives error with log of zero

[billdenney]

geom_beeswarm() is very useful, thanks for making it!

When I use it with a log transform and data including zeros, it fails. With other geoms, this would result in a point sitting on the axis (the behavior I would desire/expect). Can this be modified?

library(ggplot2)
#> Warning: package 'ggplot2' was built under R version 4.2.3
library(ggbeeswarm)

d_plot <- data.frame(A = c(0, 1), B = "B")

ggplot(d_plot, aes(x = B, y = A)) +
  geom_beeswarm()

ggplot(d_plot, aes(x = B, y = A)) +
  geom_beeswarm() +
  scale_y_log10()
#> Warning: Transformation introduced infinite values in continuous y-axis
#> Error in `geom_beeswarm()`:
#> ! Problem while computing position.
#> ℹ Error occurred in the 1st layer.
#> Caused by error in `.calculateSwarmUsingC()`:
#> ! NA/NaN/Inf in foreign function call (arg 1)
#> Backtrace:
#>      ▆
#>   1. ├─base::tryCatch(...)
#>   2. │ └─base (local) tryCatchList(expr, classes, parentenv, handlers)
#>   3. │   ├─base (local) tryCatchOne(...)
#>   4. │   │ └─base (local) doTryCatch(return(expr), name, parentenv, handler)
#>   5. │   └─base (local) tryCatchList(expr, names[-nh], parentenv, handlers[-nh])
#>   6. │     └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]])
#>   7. │       └─base (local) doTryCatch(return(expr), name, parentenv, handler)
#>   8. ├─base::withCallingHandlers(...)
#>   9. ├─base::saveRDS(...)
#>  10. ├─base::do.call(...)
#>  11. ├─base (local) `<fn>`(...)
#>  12. ├─global `<fn>`(input = base::quote("woody-kiwi_reprex.R"))
#>  13. │ └─rmarkdown::render(input, quiet = TRUE, envir = globalenv(), encoding = "UTF-8")
#>  14. │   └─knitr::knit(knit_input, knit_output, envir = envir, quiet = quiet)
#>  15. │     └─knitr:::process_file(text, output)
#>  16. │       ├─base::withCallingHandlers(...)
#>  17. │       ├─knitr:::process_group(group)
#>  18. │       └─knitr:::process_group.block(group)
#>  19. │         └─knitr:::call_block(x)
#>  20. │           └─knitr:::block_exec(params)
#>  21. │             └─knitr:::eng_r(options)
#>  22. │               ├─knitr:::in_input_dir(...)
#>  23. │               │ └─knitr:::in_dir(input_dir(), expr)
#>  24. │               └─knitr (local) evaluate(...)
#>  25. │                 └─evaluate::evaluate(...)
#>  26. │                   └─evaluate:::evaluate_call(...)
#>  27. │                     ├─evaluate (local) handle(...)
#>  28. │                     │ └─base::try(f, silent = TRUE)
#>  29. │                     │   └─base::tryCatch(...)
#>  30. │                     │     └─base (local) tryCatchList(expr, classes, parentenv, handlers)
#>  31. │                     │       └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]])
#>  32. │                     │         └─base (local) doTryCatch(return(expr), name, parentenv, handler)
#>  33. │                     ├─base::withCallingHandlers(...)
#>  34. │                     ├─base::withVisible(value_fun(ev$value, ev$visible))
#>  35. │                     └─knitr (local) value_fun(ev$value, ev$visible)
#>  36. │                       └─knitr (local) fun(x, options = options)
#>  37. │                         ├─base::withVisible(knit_print(x, ...))
#>  38. │                         ├─knitr::knit_print(x, ...)
#>  39. │                         └─knitr:::knit_print.default(x, ...)
#>  40. │                           └─evaluate (local) normal_print(x)
#>  41. │                             ├─base::print(x)
#>  42. │                             └─ggplot2:::print.ggplot(x)
#>  43. │                               ├─ggplot2::ggplot_build(x)
#>  44. │                               └─ggplot2:::ggplot_build.ggplot(x)
#>  45. │                                 └─ggplot2:::by_layer(...)
#>  46. │                                   ├─rlang::try_fetch(...)
#>  47. │                                   │ ├─base::tryCatch(...)
#>  48. │                                   │ │ └─base (local) tryCatchList(expr, classes, parentenv, handlers)
#>  49. │                                   │ │   └─base (local) tryCatchOne(expr, names, parentenv, handlers[[1L]])
#>  50. │                                   │ │     └─base (local) doTryCatch(return(expr), name, parentenv, handler)
#>  51. │                                   │ └─base::withCallingHandlers(...)
#>  52. │                                   └─ggplot2 (local) f(l = layers[[i]], d = data[[i]])
#>  53. │                                     └─l$compute_position(d, layout)
#>  54. │                                       └─ggplot2 (local) compute_position(..., self = self)
#>  55. │                                         └─self$position$compute_layer(data, params, layout)
#>  56. │                                           └─ggplot2 (local) compute_layer(..., self = self)
#>  57. │                                             └─ggplot2:::dapply(...)
#>  58. │                                               └─ggplot2 (local) apply_fun(df)
#>  59. │                                                 └─ggplot2 (local) fun(x, ...)
#>  60. │                                                   └─self$compute_panel(data = data, params = params, scales = scales)
#>  61. │                                                     └─ggbeeswarm (local) compute_panel(...)
#>  62. │                                                       └─base::lapply(...)
#>  63. │                                                         └─ggbeeswarm (local) FUN(X[[i]], ...)
#>  64. │                                                           └─beeswarm::swarmx(...)
#>  65. │                                                             └─beeswarm:::.calculateSwarmUsingC(...)
#>  66. └─base::.handleSimpleError(...)
#>  67.   └─rlang (local) h(simpleError(msg, call))
#>  68.     └─handlers[[1L]](cnd)
#>  69.       └─cli::cli_abort(...)
#>  70.         └─rlang::abort(...)

^{Created on 2023-05-19 with reprex v2.0.2}

[aroneklund]

Perhaps a solution could be to add an option to beeswarm that allows it to jitter +Inf and -Inf independently from the finite values?

beeswarm(..., allow_infinite = TRUE)

.. and I imagine that ggbeeswarm would use this option by default, and that this would give the behavior (infinite values peeking out from behind the plot borders) that ggplot2 users expect?

[billdenney]

@aroneklund , that makes sense as a possibility. I'm going to move the conversation to the beeswarm PR.