error with cross_d when a list of same-type vectors is in the .l argument

[wepelham3]

If I have a list of vectors as one of the elements in .l argument passed to cross_d, I get an error when the vectors are all of the same type.

Appears like the elements of a series of same-type vectors are getting appended together into one long vector during the process of simplifying to a data frame, which causes the error?

library(purrr)

to.cross1 <- list(x = c("apple", "banana"),
                  y = list(c("1", "2", "3"), c("a", "b")))

# error
cross_d(to.cross1)

# ------------------------------------------------------

to.cross2 <- list(x = c("apple", "banana"),
                  y = list(list("1", "2", "3"), list("a", "b")))

# working, but now y is a list of lists rather than a list of vectors
cross_d(to.cross2)

# ------------------------------------------------------

# just like first example, but with the two vectors in y being of different types
to.cross3 <- list(x = c("apple", "banana"),
                  y = list(c("1", "2", "3"), c(3.33, 5.55)))

# working, and y is a list of vectors
cross_d(to.cross3)

[lionel-]

This is a difficult problem because we're trying to be helpful and simplify lists of scalar vectors where possible. This doesn't always work out, hence the inconsistent (in appearance) results.

It feels like this problem is connected to the issue #256, i.e. we need a specification for the types of the transposed columns.

[lionel-]

We'll likely deprecate cross() and cross_df() in favour of tidyr::expand_grid(). See #768.