summarise() does not correctly coerce factors with different levels

[ghost]

I response to #1556 with the latest dev version.

I expected the output for group b-1 (2nd row) to be all 0's on the summarised variables.

library(dplyr)
df <- data.frame(grp=c("a","a","a","b","b","b"), grp2=c("2","2","2","2","1","1"), fac=c("1","1","1","1","0","0"))

dplyr::summarise(dplyr::group_by(df, grp, grp2),
                                                mean   = mean(as.numeric(fac)-1),
                                                sum1   = sum(as.numeric(fac)-1),
                                                sum2   = sum(fac == 1),
                                                any1.0 = factor(ifelse(any(fac == 1), 1, 0)),
                                                any1.1 = ifelse(any(fac == 1), 1, 0),
                                                any2.0 = factor(if(any(fac == 1)) 1 else 0),
                                                any2.1 = if(any(fac == 1)) 1 else 0,
                                                any3.0 = factor(if(any(fac == factor(1, levels = c(0,1)))) 1 else 0),
                                                any3.1 = if(any(fac == factor(1, levels = c(0,1)))) 1 else 0)

Source: local data frame [3 x 11]
Groups: grp [?]

     grp   grp2  mean  sum1  sum2 any1.0 any1.1 any2.0 any2.1 any3.0 any3.1
  (fctr) (fctr) (dbl) (dbl) (int) (fctr)  (dbl) (fctr)  (dbl) (fctr)  (dbl)
1      a      2     1     3     3      1      1      1      1      1      1
2      b      1     0     0     0      1      0      1      0      1      0
3      b      2     1     1     1      1      1      1      1      1      1

Session info -------------------------------------------------------------------------------------------------------------------------------------------------------------
 setting  value                       
 version  R version 3.2.3 (2015-12-10)
 system   x86_64, mingw32             
 ui       RStudio (0.99.875)          
 language (EN)                        
 collate  German_Germany.1252         
 tz       Europe/Berlin               
 date     2016-03-01                  

Packages -----------------------------------------------------------------------------------------------------------------------------------------------------------------
 package      * version    date       source                       
 assertthat     0.1        2013-12-06 CRAN (R 3.2.2)               
 curl           0.9.6      2016-02-17 CRAN (R 3.2.3)               
 DBI            0.3.1      2014-09-24 CRAN (R 3.2.2)               
 devtools       1.10.0     2016-01-23 CRAN (R 3.2.3)               
 digest         0.6.9      2016-01-08 CRAN (R 3.2.3)               
 dplyr        * 0.4.3.9000 2016-03-01 Github (hadley/dplyr@7d4e0ba)
 git2r          0.13.1     2015-12-10 CRAN (R 3.2.3)               
 httr           1.1.0      2016-01-28 CRAN (R 3.2.3)               
 knitr          1.12.3     2016-01-22 CRAN (R 3.2.3)               
 lazyeval       0.1.10     2015-01-02 CRAN (R 3.2.2)               
 magrittr       1.5        2014-11-22 CRAN (R 3.2.2)               
 memoise        1.0.0      2016-01-29 CRAN (R 3.2.3)               
 nycflights13 * 0.1        2014-07-22 CRAN (R 3.2.3)               
 R6             2.1.2      2016-01-26 CRAN (R 3.2.3)               
 Rcpp           0.12.3     2016-01-10 CRAN (R 3.2.3)               
 rstudioapi     0.5        2016-01-24 CRAN (R 3.2.3)               
 withr          1.0.1      2016-02-04 CRAN (R 3.2.3)        

[1va]

You could specify the levels (of the if/else results) to prevent wrong/unexpected concatenation of factors with different levels:
factor(ifelse(any(fac == 1), 1, 0), levels = c(0, 1))

But I agree that the expected behaviour would be that the levels are handled automatically similar to
unlist(list(factor(0), factor(1))

Compare c(factor(0), factor(1)) and c(factor(0, levels = c(0, 1)), factor(1, levels = c(0, 1)))

[hadley]

Minimal reprex:

data_frame(x = 1:2) %>% 
  group_by(x) %>% 
  summarise(
    y = if(x == 1) "a" else "b",
    z = factor(y)
  )