Failure to print a tibble with an S4 column

[hplieninger]

When I want to print a tibble containing an S4 column, I get an error with tibble 1.4.1 as well as with the current GitHub version, but not with version 1.3.4.

library(tibble)
#> Warning: package 'tibble' was built under R version 3.4.3
a1 <- nycflights13::flights[, c("flight", "air_time")]
a1$duration <- lubridate::dminutes(a1$air_time)
str(a1)
#> Classes 'tbl_df', 'tbl' and 'data.frame':    336776 obs. of  3 variables:
#>  $ flight  : int  1545 1714 1141 725 461 1696 507 5708 79 301 ...
#>  $ air_time: num  227 227 160 183 116 150 158 53 140 138 ...
#>  $ duration:Formal class 'Duration' [package "lubridate"] with 1 slot
#>   .. ..@ .Data: num  13620 13620 9600 10980 6960 ...
print(a1)
#> Error: Incompatible duration classes (Duration, numeric). Please coerce with `as.duration`.

[ghost]

I posted a superset of the following on StackOverflow in trying to get to the bottom of this issue (https://stackoverflow.com/questions/48087358/tibbles-reject-lubridates-duration-and-period-objects). It may be useful - and the reprexes (reprices?) are more minimal.

The following code successfully produces a duration object.

lubridate::as.duration(1)
#> "1s"

However, when I try to print tibbles using a duration object, I get an error message.

tibble::tibble(y = lubridate::as.duration(1))
#> Error: Incompatible duration classes (Duration, numeric). Please coerce with `as.duration`.

Replacing tibble::tibble with base::data.frame works.

data.frame(y = lubridate::as.duration(1))
#>   y
#> 1 1s

Using tibble::as_tibble to coerce these data.frames to tibbles yields the same error as before.

tibble::as_tibble(data.frame(y = lubridate::as.duration(1)))
#> Error: Incompatible duration classes (Duration, numeric). Please coerce with `as.duration`.

Digging in the source code, I find the following chain of dependencies that give the same error message: as_tibble.data.frame --> list_to_tibble --> new_tibble

In tibble:::list_to_tibble, the only argument passed to tibble::new_tibble is x. Consequently, subclass is assigned the default value of NULL, and the penultimate line of tibble::new_tibble becomes

class(x) <- c("tbl_df", "tbl", "data.frame")

The objects have a structure, but trying to print them yields an error.

x <- data.frame(y = lubridate::as.duration(1))
class(x) <- c("tbl_df", "tbl", "data.frame")
str(x)
#> Classes ‘tbl_df’, ‘tbl’ and 'data.frame':	1 obs. of  1 variable:
#>  $ x:Formal class 'Duration' [package "lubridate"] with 1 slot
#>   .. ..@ .Data: num 1
x
#> Error: Incompatible duration classes (Duration, numeric). Please coerce with `as.duration`.

Consequently, it appears that assigning the class of the data.frame x the vector c("tbl_df", "tbl", "data.frame") causes R to try to coerce x in a way that throws an error when x is printed.

Furthermore, given that tibble::tibble also calls tibble::as_tibble (albeit not on a data.frame), I would hazard a guess that my issues with tibble::tibble have the same cause.

Package versions

• Tibble: 1.4.1
• Lubridate: 1.7.1
• R: 3.4.3

[krlmlr]

Thanks for the analysis. It looks like creating the tibble succeeds, only printing fails:

x <- tibble::tibble(y = lubridate::as.duration(1))
print(x)
## Error: ...

The source of the error seems to be the pillar() constructor:

x <- pillar::pillar(lubridate::as.duration(1))
#> Error: Incompatible duration classes (Duration, numeric). Please coerce with `as.duration`.

Created on 2018-01-14 by the reprex package (v0.1.1.9000).

Closing in favor of r-lib/pillar#88.

[jschelbert]

Hi,
It seems that the same issue also occurs for intervals (with a slightly different error message).
Not sure this is also due to the pillar bug or something else...

library(tidyverse)
#> -- Attaching packages -------------------------------------------------------------------------------------------------------------------- tidyverse 1.2.1 --
#> v ggplot2 2.2.1     v purrr   0.2.4
#> v tibble  1.4.2     v dplyr   0.7.4
#> v tidyr   0.7.2     v stringr 1.2.0
#> v readr   1.1.1     v forcats 0.2.0
#> -- Conflicts ----------------------------------------------------------------------------------------------------------------------- tidyverse_conflicts() --
#> x dplyr::filter() masks stats::filter()
#> x dplyr::lag()    masks stats::lag()
library(lubridate)
#> 
#> Attache Paket: 'lubridate'
#> The following object is masked from 'package:base':
#> 
#>     date

df <- data.frame(begin = c(ymd_hm("2018-01-01 00:00"),
                           ymd_hm("1900-01-01 00:00"),
                           ymd_hm("1900-01-01 00:00")), 
                 end = c(ymd_hm("2018-01-04 00:00"),
                         ymd_hm("3000-01-01 00:00"),
                         ymd_hm("3000-12-31 00:00")))

df %>% mutate(test = interval(start=begin, end=end))
#>                 begin                 end
#> 1 2018-01-01 01:00:00 2018-01-04 01:00:00
#> 2 1900-01-01 01:00:00 3000-01-01 01:00:00
#> 3 1900-01-01 01:00:00 3000-12-31 01:00:00
#>                                               test
#> 1 2018-01-01 01:00:00 CET--2018-01-04 01:00:00 CET
#> 2 1900-01-01 01:00:00 CET--3000-01-01 01:00:00 CET
#> 3 1900-01-01 01:00:00 CET--3000-12-31 01:00:00 CET

df <- tibble(begin = c(ymd_hm("2018-01-01 00:00"),
                           ymd_hm("1900-01-01 00:00"),
                           ymd_hm("1900-01-01 00:00")), 
                 end = c(ymd_hm("2018-01-04 00:00"),
                         ymd_hm("3000-01-01 00:00"),
                         ymd_hm("3000-12-31 00:00")))

df %>% mutate(test = interval(start=begin, end=end))
#> Note: method with signature 'Interval#ANY' chosen for function '-',
#>  target signature 'Interval#Interval'.
#>  "ANY#Interval" would also be valid
#> Error in round_x - lhs: Arithmetic operators undefined for 'Interval' and 'Interval' classes:
#>   convert one to numeric or a matching time-span class.

[ghost]

@jschelbert

The following code yields the same error, so your error is also due to the pillar bug:

pillar::pillar(lubridate::interval(lubridate::ymd("2017-01-01"), lubridate::ymd("2018-01-01")))

[krlmlr]

Printing S4 objects should work with the latest version of pillar.

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.