Unnesting tibble with zero rows removes unnested column

[frederikziebell]

Here is an example that produces the problem:

df <- tibble(
 a = list(c(1, 2), 3),
 b = c(4,5)
)

df %>%
  unnest(a) %>%
  colnames
# [1] "b" "a"

df %>%
  filter(b==1) %>%
  unnest(a) %>%
  colnames
# [1] "b"

If the special case occurs that the tibble being unnested has no rows, all code acting on the unnested column breaks.

[krlmlr]

Thanks, confirmed with both the CRAN version and 581c2c0.

[hadley]

I see an error (possible due to dev dplyr)

library(tidyr)

df <- tibble::tibble(
 a = list(c(1, 2), 3),
 b = c(4,5)
)

df %>%
  unnest(a) %>%
  names()
#> [1] "b" "a"

df %>%
  filter(b==1) %>%
  unnest(a) %>%
  names()
#> Error in data.matrix(data): (list) object cannot be coerced to type 'double'

^{Created on 2019-01-04 by the reprex package (v0.2.1.9000)}

[frederikziebell]

I get the same error. Here is my sessionInfo:

R version 3.5.0 (2018-04-23)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=German_Germany.1252  LC_CTYPE=German_Germany.1252    LC_MONETARY=German_Germany.1252 LC_NUMERIC=C                   
[5] LC_TIME=German_Germany.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] tidyr_0.8.2

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.19     crayon_1.3.4     dplyr_0.7.7      assertthat_0.2.0 R6_2.3.0         magrittr_1.5     pillar_1.3.0     rlang_0.3.0.1   
 [9] rstudioapi_0.8   bindrcpp_0.2.2   tools_3.5.0      glue_1.3.0       purrr_0.2.5      yaml_2.2.0       compiler_3.5.0   pkgconfig_2.0.2 
[17] bindr_0.1.1      knitr_1.20       tidyselect_0.2.5 tibble_1.4.2  

[hadley]

But I see the same basic problem with a different example:

library(tidyr)
data.frame(a = character(), b = character()) %>% 
  unnest(b) %>% 
  names()
#> [1] "a"

^{Created on 2019-01-04 by the reprex package (v0.2.1.9000)}

[khailper]

I'll try this.

[hadley]

A bit more of a reprex to help me understand the problem:

library(tidyr)

df <- tibble::tibble(a = list(1), y = 1)
df %>% unnest(a) %>% str()
#> Classes 'tbl_df', 'tbl' and 'data.frame':    1 obs. of  2 variables:
#>  $ y: num 1
#>  $ a: num 1
df %>% .[0, ] %>% unnest(a) %>% str()
#> Classes 'tbl_df', 'tbl' and 'data.frame':    0 obs. of  1 variable:
#>  $ y: num

^{Created on 2019-01-22 by the reprex package (v0.2.1.9000)}

[hadley]

I think the root cause is actually this:

list_col_type(character())
#> [1] "data.frame"