Parsing the x= and y= arguments is messy. This is due to the evolution of the function from pre-rlang/tidyeval to the more friendly tidy evaluation syntax.
I think I'd like to deprecate accepting string names for y= and x=. The issue arises with complex outcomes, like time to event: Surv(ttdeath, death), which is tricky to manage when you accept string versions and unquoted versions.
Parsing the
x=
andy=
arguments is messy. This is due to the evolution of the function from pre-rlang/tidyeval to the more friendly tidy evaluation syntax.I think I'd like to deprecate accepting string names for
y=
andx=
. The issue arises with complex outcomes, like time to event:Surv(ttdeath, death)
, which is tricky to manage when you accept string versions and unquoted versions.