Improved stat_function() documentation #3362

[thiyangt]

Closes #3362

[lionel-]

Could you redocument with devtools::document() (or from the Build menu in RStudio) please? This will regenerate the .Rd files which contain the documentation. The Rd changes should be part of the commit.

[lionel-]

Would you like to also add a warning when mapping or data is supplied, as suggested in #3362 (comment)?

[lionel-]

Thanks! The warnings will also need unit tests that you can add in tests/testthat/test-stats-function.r. This should look like:

expect_warning(stat_function(aes(), fun = identity), "Ignoring `mapping`")

However I see the warning currently is:

warn_overwritten_args("stat_function()", "mapping", "group")
#> Warning message:
#> stat_function(): Ignoring `mapping` because `group` was provided.

Which doesn't look right. It's probably better to just call warning() directly with custom messages. Any suggestion as to how best formulate the two warnings @paleolimbot? I'm not sure I entirely understand its semantics.

[paleolimbot]

You're right @lionel- ...I think it would be more accurate here to write

   warning("`mapping` is not used by stat_function()", call. = FALSE)

[lionel-]

The PR just needed another small change, the warnings needed to be above the layer() call, otherwise the stat_function() would not return anything. No big deal, I have pushed a quick fix: 6ba9275.

Thanks for these improvements to ggplot2!

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/