all.equal.tbl_df should return TRUE for tbl_df with Date column

[jullybobble]

a is a data.frame with a Date column.

a <- data.frame(date = as.Date("2015-06-07"))

When serialised and deserialised with readr's write_csv and read_csv, the resulting data.frame b_tbl is equal to a according to base:::all.equal.default, but not according to dplyr:::all.equal.tbl_df:

b_tbl <- readr::read_csv(readr::write_csv(a, path = ""))
b <- dplyr:::as.data.frame.tbl_df(b_tbl)
a_tbl <- dplyr:::as.tbl(a)

dplyr:::all.equal.tbl_df(a_tbl, b_tbl)
# "Incompatible type for column date: x Date, y Date"

base:::all.equal.default(a_tbl, b_tbl) 
# TRUE

base:::all.equal.default(a, b)
 # TRUE

dplyr:::all.equal.tbl_df(a, b)
# "Incompatible type for column date: x Date, y Date"

I would expect all 4 all.equal statements above to return TRUE.

Note that this occurs only if b_tbl is the result of a write_csv and read_csv, so I am not sure whether the issue is related to readr or dplyr, but the inconsistency shows in dplyr:

dplyr:::all.equal.tbl_df(a_tbl, a)
# TRUE

[romainfrancois]

Hmm. The issue is that for some reason the Date in a is encoded as a numeric instead of an integer :

> .Internal(inspect(( a_tbl$date )))
@7f87e3e990c8 14 REALSXP g0c1 [OBJ,MARK,NAM(2),ATT] (len=1, tl=0) 16593
ATTRIB:
  @7f87e3a3db48 02 LISTSXP g0c0 [MARK]
    TAG: @7f87f2006148 01 SYMSXP g1c0 [MARK,NAM(2),LCK,gp=0x4000] "class" (has value)
    @7f87e3e99098 16 STRSXP g0c1 [MARK,NAM(2)] (len=1, tl=0)
      @7f87f21f18a8 09 CHARSXP g1c1 [MARK,gp=0x61] [ASCII] [cached] "Date"
> .Internal(inspect(( b_tbl$date )))
@7f87f6e20cb8 13 INTSXP g0c1 [OBJ,NAM(2),ATT] (len=1, tl=0) 16593
ATTRIB:
  @7f87f3b181f8 02 LISTSXP g0c0 []
    TAG: @7f87f2006148 01 SYMSXP g1c0 [MARK,NAM(2),LCK,gp=0x4000] "class" (has value)
    @7f87f6e20d18 16 STRSXP g0c1 [NAM(2)] (len=1, tl=0)
      @7f87f21f18a8 09 CHARSXP g1c1 [MARK,gp=0x61] [ASCII] [cached] "Date"

[hadley]

@romainfrancois sometimes dates are numerics and sometimes they're integers :/

[romainfrancois]

Yes apparently. I'll come up with a workaround.