modify() not maintaining class for atomic vectors

[MrFlick]

The documentation for modify says that the .x parameter is "A list or atomic vector." and the "Value" section says that the functions will return "An object the same class as .x" However for atomic vectors, it seems to return a list, and not a vector of the same class as .x.

x <- c("a","b","c","d","e")
y <- c(1, 2, 3, 4, 5)
z <- c(T, F, T, F, T)

class(modify(x,function(x) "k"))
# [1] "list"      ## not class(x)

class(modify(y,function(x) 10))
# [1] "list"     ## not class(y)

class(modify_if(x, z, function(x) "k"))
# [1] "list"     ## not class(x)

class(modify_if(y, z, function(x) x+10))
# [1] "list"     ## not class(y)

The help page also says that modify() is basically the same as x[] <- map(x, .f) and ifs that's the case i'd expect map() to return a list, But even on the ?map help page it calls out: "See the modify() family for versions that return an object of the same type as the input."

Tested with purrr_0.2.4 and R version 3.4.1 (2017-06-30)

[lionel-]

Should be fixed by #463

[stufield]

Question: has this issue been resolved? I stubbed my toe on it today using the CRAN version purrr_0.2.5 and the list conversion is still happening. I reverted to the dev version, which appears to be purrr_0.2.4.9000, and it's fixed there. My confusion arises from my assumption that 0.2.5 comes after 0.2.4.9000 in the version schema. Am I wrong? Or worse, was it fixed, then reverted? Or should the development version actually read 0.2.5.9000. In which case I humbly withdraw my question ;)

[yutannihilation]

It seems some commits included in v0.2.5 are forgot to be merged into master...

v0.2.4...v0.2.5