purrr:contains and dplyr::contains is likely to cause confusion

[sebastiansauer]

Both purrr and dplyr have a contains() function, but with different syntax.

For example, if I load dplyr first, and purr afterwards, it will bite me:

   library(dplyr)
   library(purrr)
   data(tips, package = "reshape2")
   tips %>% select(contains("total"))
   ## Error in .p(.x[[i]], ...) : argument ".y" is missing, with no default

May it would be helpful to disambiguate the two functions?

[hadley]

@lionel- any thoughts on a better name? We might need to consider more prefixes for purrr functions since it's going to be loaded increasingly commonly

[lionel-]

we could prefix with vec_ secondary functions that operate on vectors, generic and atomic? But then what would be the prefix for atomic vectors functions?

[hadley]

atm_? But if you think of contains() as a function that operates on vectors, maybe it should go in vctrs? I think we'll need to figure out what goes where.

[lionel-]

this makes me think that the tidyverse is starting to look like Emacs lisp functions ;) Elisp doesn't have namespaces so all function names have to be prefixed.

maybe it should go in vctrs?

I'm not sure, this is a wrapper around some(). We'll also have to figure out what should be the dependency graph of those packages.

atm_?

ah nice, I was stuck with ato_ for some reason.

[DataStrategist]

Or just change one of the verbs to has()?

[jtr13]

Yes, confusion caused.

> select(x100, contains("P"))
Error in .p(.x[[i]], ...) : argument ".y" is missing, with no default

[hadley]

And similarly for order_by()

[kleinschmidt]

Why did sort_by need to be removed as well, and what's the recommended replacement (other than "use tibbles")?