Proper syntax for character vector quasi-quotation

[mjsteinbaugh]

Hi all, I'm reading through the new dplyr programming vignette and am working on updating the syntax used in my packages. I'm confused how to properly pass in character vectors (e.g. for select(), arrange()) with the new tidyeval methods using standard evaluation. What's the recommended syntax for quasi-quotation of a character vector for column name matching in place of the now deprecated .dots method? Everything else is working great for me in the latest release candidate.

head(starwars)
cols <- c("name", "height", "mass")

# `select()` examples ====
# Deprecated dplyr 0.5.0 method
select_(starwars, .dots = cols)

# Works, but with integer positions
quosure <- quo(1:3)
select(starwars, !!quosure)

# Works, but only for single name
quosure <- quo(name)
select(starwars, !!quosure)

# Works, but is there an unquote approach like above?
select(starwars, one_of(cols))

# `arrange()` examples ====
# Deprecated dplyr 0.5.0 method
arrange_(starwars, .dots = cols)
# Current best practice SE method for dplyr 0.6.0?

Thanks, Mike

[lionel-]

this would be !!! rlang::syms(chr_vector). @hadley should sym() and syms() be reexported in dplyr?

[mjsteinbaugh]

Awesome, works great. I was close in the rlang documentation!

select(starwars, !!!rlang::syms(cols))
arrange(starwars, !!!rlang::syms(cols))

[hadley]

I don't think we need to re-export it because you're using the

[MikeBadescu]

For completion, is there other solution than using !!! rlang::syms(chr_vector)?

[lionel-]

syms(x) is equivalent to lapply(x, as.name)

[MikeBadescu]

Lionel, thank you! In general, it would help if SE would be first class throughout in tidyverse.