arrange() on a data frame causes R to crash

[wch]

If arrange is called with no ... arguments, it causes R to crash.

library(dplyr)
arrange(mtcars)
#  *** caught segfault ***
# address 0xfffffffffffffff8, cause 'memory not mapped'

[romainfrancois]

I get a different problem :

> arrange(mtcars)
Erreur : not a promise

[romainfrancois]

@hadley I've made it so arrange returns its input when there is no additional parameters, so no arranging at all. Is this alright ?

[hadley]

That's right. The only function that shouldn't do that is maybe summarise()

[romainfrancois]

For a grouped data frame, summarise returns the labels attribute, which sounds good enough. For a non grouped, we get an empty data frame.

> .Internal(inspect(summarise(mtcars))
+ )
@1081290a0 19 VECSXP g0c0 [OBJ,ATT] (len=0, tl=0)
ATTRIB:
  @108128e00 02 LISTSXP g0c0 []
    TAG: @100823548 01 SYMSXP g1c0 [MARK,NAM(2),LCK,gp=0x4000] "class" (has value)
    @108124d68 16 STRSXP g0c1 [] (len=1, tl=0)
      @1008a7e60 09 CHARSXP g1c2 [MARK,gp=0x61,ATT] [ASCII] [cached] "data.frame"
    TAG: @100823078 01 SYMSXP g1c0 [MARK,NAM(2),LCK,gp=0x4000] "names" (has value)
    @108127390 16 STRSXP g0c0 [] (len=0, tl=0)
    TAG: @10082d060 01 SYMSXP g1c0 [MARK,LCK,gp=0x4000] "row.names" (has value)
    @108124e28 13 INTSXP g0c1 [] (len=2, tl=0) -2147483648,-1

But this data frames pretends to have one row and zero columns,
Maybe it could be simpler and just be equivalent to what data.frame() gives, so row.names would be an empty character vector instead.

[hadley]

Yeah, that sounds like a good idea.