Tangentially related, but I wonder whetherrlang::data_sym() passingNULL through could resolve some of this headache:

rlang::data_sym(NULL)#> NULL

Instead of its current behavior:

rlang::data_sym(NULL)#> Error in `sym()`:#> ! Can't convert `NULL` to a symbol.

Especially because I feel like this "optional character-input" problem has already been solved in tidyselect (withall_of()/any_of()), so it'd be nice ifdata_sym()/data_syms() could let us get something close to tidyselect semantics within tidyeval contexts (likeaes()) by lettingNULL pass through to mean an empty (NULL) selection. Ex:

df<-data.frame(x=1)# Tidyselect behaviortidyselect_get<-function(df,col) {tidyselect::eval_select(tidyselect::all_of(col),df)}tidyselect_get(df,"x")#> x#> 1tidyselect_get(df,NULL)#> named integer(0)# Tidyeval with `data_sym()`tidyeval_get<-function(df,col) {rlang::eval_tidy(rlang::data_sym(col),df)}tidyeval_get(df,"x")#> [1] 1tidyeval_get(df,NULL)#> Error in `sym()`:#> ! Can't convert `NULL` to a symbol.# Tidyeval with modified `data_sym2()` that passes NULL throughdata_sym2<- \(x)if (!is.null(x))rlang::data_sym(x)tidyeval_get2<-function(df,col) {rlang::eval_tidy(data_sym2(col),df)}tidyeval_get2(df,"x")#> [1] 1tidyeval_get2(df,NULL)#> NULL

Thanks June, your comments make a lot of sense to me. I'd agree that this havingdata_sym() acceptNULL would be the most elegant. I don't think ggplot2 should provide a user-exposed wrapper for this, as we'd have to maintain this into perpetuity. Rather, we could petition {rlang} if they'd be interested in this change. That allows us to include this PR as current advice and update the advice if and once we can usedata_sym(NULL).

Rather, we could petition {rlang} if they'd be interested in this change.

Yeah, that's the intended build-up of my comment! Would be nice for the FR to be backed by a need in ggplot 😄

I'm personally partial to advocating embrace for people wanting to build ggplot2 wrapper functions. The issue noted in the original issue (name of column coming as a string) is easily solved:

plot_embrace<-function(dat,xvar,yvar,colorvar=NULL,shapevar=NULL) {myplot<- ggplot(dat,aes(x={{xvar }},y={{yvar }},color={{colorvar }},shape={{shapevar }}))+    geom_point()return(myplot)}whichvar<- sample(c('sex','employment'),1)plot_embrace(plotdat,weight,height,colorvar=age,shapevar=!!sym(whichvar))

I'm happy to recommend the!!sym() trick, but it'd be a different approach than this PR, so I'll create a new one.