Column silently dropped when its name conflicts with value col name in gather.data.frame()

[jarodmeng]

gather.data.frame() computes which columns to gather using this snippet.

  quos <- quos(...)
  if (is_empty(quos)) {
    gather_vars <- setdiff(names(data), c(key_var, value_var))
  } else {
    gather_vars <- unname(tidyselect::vars_select(names(data), !!! quos))
  }

When the user doesn't provide a ... argument, the gather columns are computed to be anything that's not key or value. This logic works when there's no existing column having the same name as value. If there is, that column would be silently dropped from the gathering.

I've produced a simple reprex below to illustrate.

packageVersion("tidyr")
#> [1] '0.7.0'

library(tidyr)
library(tibble)

XYZ <- data.frame(
  X = rnorm(2, 0, 1),
  Y = rnorm(2, 0, 2),
  Z = rnorm(2, 0, 4)
)

gather(XYZ, name, value)
#>   name      value
#> 1    X  0.1051179
#> 2    X  1.4617993
#> 3    Y -0.9121167
#> 4    Y  2.5378557
#> 5    Z -2.1496032
#> 6    Z -2.4815070

gather(XYZ, name, Y)
#>            Y name          Y
#> 1 -0.9121167    X  0.1051179
#> 2  2.5378557    X  1.4617993
#> 3 -0.9121167    Z -2.1496032
#> 4  2.5378557    Z -2.4815070

[hadley]

Slightly more minimal reprex, along with behaviour for tibbles.

library(tidyr)

XYZ <- data.frame(
  X = 1,
  Y = 1,
  Z = 2
)

XYZ %>% gather(key = "name", value = "Y")
#>   Y name Y
#> 1 1    X 1
#> 2 1    Z 2
XYZ %>% tibble::as.tibble() %>% gather(key = "name", value = "Y")
#> Error: Column `Y` must have a unique name

Created on 2017-11-16 by the reprex package (0.1.1.9000).

[hadley]

Basically same issue as #255 but for gather(), which uses a slightly different code path.