mutate errors when coercing to POSIXlt types

[jimhester]

It happens if you use as.POSIXlt() or strptime() directly.

data.frame(time='2014/01/01 10:10:10') %>% dplyr::mutate(time=as.POSIXlt(time))
# Error: wrong result size (11), expected 1 or 1

You get the same error wrong result size (11) regardless of the length of the data.frame
The expected output would be the same as using plyr::mutate()

str(data.frame(time='2014/01/01 10:10:10') %>% plyr::mutate(time=as.POSIXlt(time)))
# 'data.frame': 1 obs. of  1 variable:
#  $ time: POSIXlt, format: "2014-01-01 10:10:10"

This error occurs in both dplyr version 0.3.0.1 on CRAN and latest github versions.

Session Info:

Session info------------------------------------------
 setting  value
 version  R version 3.1.1 (2014-07-10)
 system   x86_64, darwin13.3.0
 ui       X11
 language (EN)
 collate  en_US.UTF-8
 tz       America/New_York

Packages---------------------------------------------
 package    * version    date       source
 assertthat   0.1        2013-12-06 CRAN (R 3.1.1)
 colorspace   1.2.4      2013-09-30 CRAN (R 3.1.1)
 crayon       1.0.0      2014-09-28 Github (gaborcsardi/crayon@be81a40)
 DBI          0.3.0      2014-09-03 CRAN (R 3.1.1)
 devtools   * 1.6.0.9000 2014-09-25 Github (hadley/devtools@51ebc3a)
 digest       0.6.4      2013-12-03 CRAN (R 3.1.1)
 dplyr      * 0.3.0.9000 2014-10-08 Github (hadley/dplyr@87ffc2d)
 evaluate     0.5.5      2014-04-29 CRAN (R 3.1.1)
 formatR      1.0        2014-08-25 CRAN (R 3.1.1)
 ggplot2    * 1.0.0      2014-05-21 CRAN (R 3.1.1)
 gtable       0.1.2      2012-12-05 CRAN (R 3.1.1)
 htmltools    0.2.6      2014-09-08 CRAN (R 3.1.1)
 httr         0.5        2014-09-02 CRAN (R 3.1.1)
 knitr        1.6.21     2014-10-07 Github (yihui/knitr@513d68f)
 lazyeval     0.1.9      2014-10-03 Github (hadley/lazyeval@6dcbee9)
 magrittr     1.0.1      2014-05-15 CRAN (R 3.1.1)
 MASS         7.3.33     2014-05-05 CRAN (R 3.1.1)
 munsell      0.4.2      2013-07-11 CRAN (R 3.1.1)
 plyr         1.8.1      2014-02-26 CRAN (R 3.1.1)
 proto        0.3.10     2012-12-22 CRAN (R 3.1.1)
 R6           2.0        2014-08-19 CRAN (R 3.1.1)
 Rcpp         0.11.3     2014-09-29 CRAN (R 3.1.1)
 RCurl        1.95.4.3   2014-07-29 CRAN (R 3.1.1)
 reshape2     1.4        2014-04-23 CRAN (R 3.1.1)
 rex        * 0.1.0      <NA>       local
 rmarkdown    0.3.3      2014-09-17 CRAN (R 3.1.1)
 roxygen2     4.0.2.9000 2014-10-01 Github (klutometis/roxygen@78d5d7b)
 rstudioapi   0.1        2014-03-27 CRAN (R 3.1.1)
 scales       0.2.4      2014-04-22 CRAN (R 3.1.1)
 setwidth   * 1.0.3      2013-01-16 CRAN (R 3.1.1)
 stringr      0.6.2      2012-12-06 CRAN (R 3.1.1)
 testthat   * 0.9.0.9000 2014-09-28 Github (hadley/testthat@1e64e0c)
 vimcom     * 1.0.0      2014-09-18 Github (jalvesaq/VimCom@70fec5c)
 yaml         2.1.13     2014-06-12 CRAN (R 3.1.1)

[hadley]

I don't think we can support POSIXlt in data frames - it really is a list of 11 components. You're much better off with POSIXct.

[romainfrancois]

We could, with some work. I think what I'm going to do is use POSIXlt as an advanced example for #650 if it still does make sense when I tackle it.

[jimhester]

Fair enough, I didn't realize how POSIXlt was implemented. POSIXct works fine for my use case as well.

[romainfrancois]

I've made this explicit:

> data.frame(time='2014/01/01 10:10:10') %>% dplyr::mutate(time=as.POSIXlt(time))
Erreur : `mutate` does not support `POSIXlt` results

Perhaps not as good as actually supporting POSIXlt, but less it'd be more work.