Use the fully qualified forms for quos()#3161
Conversation
lionel-
left a comment
There was a problem hiding this comment.
Our default practice is to import the whole rlang package and use the non-qualified form. However this PR makes calls to tidy eval function more consistent with other calls to rlang since rlang is not imported in ggplot.
|
@lionel- Thanks, I'm merging this for consistency this time. Do you think ggplot2 should also import the whole rlang package? If it's a common practice among tidyverse, ggplot2 should follow, and using unqualified form seems a bit better in terms of the speed. |
|
I don't have a strong opinion about this. Maybe best discussed with the regular contributors? |
|
Sure. |
|
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/ |
Though
quos()are imported inutilities-tidy-eval.R, I think it's good practice to stick with the fully qualified forms in package code (I'm tolerant for test code). There are a few places that lacksrlang::, so this PR added them.In addition, I want to remove this line to make it clear that we don't encourange the unqualified forms of
quo()andquos(). I guess this was added by mistake.ggplot2/R/ggplot2.r
Line 6 in 033fb52