Unhelpful error message when using dplyr::between() in join_by()

[mgirlich]

I accidentally used dplyr::between() instead of between() in join_by(). Not sure whether you want to accept it with a namespace or not but the error message isn't very helpful.

dplyr::join_by(dplyr::between(y, a, b))
#> Error in `as_string()`:
#> ! Can't convert a call to a string.
#> Backtrace:
#>     ▆
#>  1. └─dplyr::join_by(dplyr::between(y, a, b))
#>  2.   └─dplyr:::parse_join_by_expr(exprs[[i]], i, error_call = error_call)
#>  3.     └─rlang::as_string(expr[[1]])
#>  4.       └─rlang:::abort_coercion(x, "a string")
#>  5.         └─rlang::abort(msg, call = call)

^{Created on 2023-05-02 with reprex v2.0.2}

[DavisVaughan]

@hadley it probably makes sense to attempt to support dplyr:: namespacing right?

[hadley]

Yeah, probably.

[balthasars]

I accidentally used dplyr::between() instead of between() in join_by(). Not sure whether you want to accept it with a namespace or not but the error message isn't very helpful.

dplyr::join_by(dplyr::between(y, a, b))
#> Error in `as_string()`:
#> ! Can't convert a call to a string.
#> Backtrace:
#>     ▆
#>  1. └─dplyr::join_by(dplyr::between(y, a, b))
#>  2.   └─dplyr:::parse_join_by_expr(exprs[[i]], i, error_call = error_call)
#>  3.     └─rlang::as_string(expr[[1]])
#>  4.       └─rlang:::abort_coercion(x, "a string")
#>  5.         └─rlang::abort(msg, call = call)

Created on 2023-05-02 with reprex v2.0.2

Would be super neat if dplyr would support that so the new join_by() feature can also be used in situations where calling dplyr::between() is the only safe option. Thanks for the great work on dplyr!

[DavisVaughan]

What do you mean by safe?

[balthasars]

What do you mean by safe?

Safe as in not loading the entire dplyr namespace using library() in legacy code projects that have the potential of namespace conflicts.

I suppose a potential workaround would be to use library(dplyr, include.only = c("between")), which sort of invalidates my argument.

[DavisVaughan]

join_by() interprets its input in a very special way, so if you don't load dplyr you can still call between() without namespacing it, i.e.

dplyr::join_by(between(x, y, z))
#> Join By:
#> - between(x, y, z)

^{Created on 2023-09-18 with reprex v2.0.2}

[balthasars]

join_by() interprets its input in a very special way, so if you don't load dplyr you can still call between() without namespacing it, i.e.

dplyr::join_by(between(x, y, z))
#> Join By:
#> - between(x, y, z)

Created on 2023-09-18 with reprex v2.0.2

😮 I didn't even consider that, very cool! Thank you for providing this illuminating example! And thanks for the new join_by() feature, I've become an avid user 👍🏼