[BUGZILLA #16648] complete.cases() give misleading message

[MichaelChirico]

I have a data frame:

df

    Time Treat

1 2015-03-28 1
2 2015-03-28 1
3 2015-03-28 1

dput(df)

structure(list(Time = structure(list(sec = c(0, 0, 0), min = c(0L,
0L, 0L), hour = c(0L, 0L, 0L), mday = c(28L, 28L, 28L), mon = c(2L,
2L, 2L), year = c(115L, 115L, 115L), wday = c(6L, 6L, 6L), yday = c(86L,
86L, 86L), isdst = c(0L, 0L, 0L), zone = c("CST", "CST", "CST"
), gmtoff = c(NA_integer_, NA_integer_, NA_integer_)), .Names = c("sec",
"min", "hour", "mday", "mon", "year", "wday", "yday", "isdst",
"zone", "gmtoff"), class = c("POSIXlt", "POSIXt")), Treat = c(1L,
1L, 1L)), .Names = c("Time", "Treat"), row.names = c(NA, 3L), class = "data.frame")

when I apply complete.cases() on it, it gives error message:

complete.cases(df)

Error in complete.cases(df) : not all arguments have the same length

It seems that complete.case() cannot handle the first column. I think it should give a warning message such as the class/type is not supported, instead of error message without any hints about the error.

version

           _                           

platform x86_64-w64-mingw32
arch x86_64
os mingw32
system x86_64, mingw32
status
major 3
minor 2.3
year 2015
month 12
day 10
svn rev 69752
language R
version.string R version 3.2.3 (2015-12-10)
nickname Wooden Christmas-Tree

---

METADATA

• Bug author - Jinsong Zhao
• Creation time - 2015-12-30 01:01:58 UTC
• Bugzilla link
• Status - UNCONFIRMED
• Alias - None
• Component - Misc
• Version - R 3.2.2
• Hardware - Other Other
• Importance - P5 enhancement
• Assignee - R-core
• URL -
• Modification time - 2016-01-03 21:30 UTC

[MichaelChirico]

The problem here is that complete.cases doesn't use the R function length() to determine the length of the Time column, it uses the C macro LENGTH, which ignores the fact that Time is of class POSIXlt, and sees it as a length 11 list.

A workaround is to convert the time to POSIXct instead of POSIXlt.

It looks quite messy to fix this properly, so I'll just add a note to the documentation warning about the problem.

---

METADATA

• Comment author - Duncan Murdoch
• Timestamp - 2016-01-02 16:01:54 UTC

[MichaelChirico]

It actually looks pretty straightforward to translate the C code to R code which will pay attention to the classes. If that doesn't slow it down too much, I'll make that change.

---

METADATA

• Comment author - Duncan Murdoch
• Timestamp - 2016-01-03 21:30:41 UTC