[R] sort() for time/date objects (was: more on date conversion differences in 2.2.1 vs 2.3.1)

[Achim Zeileis]

On Sat, 12 Aug 2006 09:25:10 +1000 paul sorenson wrote:

> With dates I get different results with 2.2.1 and 2.3.1.  From my 
> somewhat naive point point of view, the 2.2.1 behaviour seems more
> sensible.

Your example below does not really make the source of the problem
obvious. I tried to see where the problem comes from and the following
seems to happen:
  xtabs() calls factor() calls sort(unique()) for computing the levels
And in the last step sort() seems to strip off the "class" attribute,
leaving only the underlying numeric vector.

Of course, the functions never claimed to work for "POSIXct" objects,
hence you should call xtabs() appropriately, e.g. via
  xtabs(V2 ~ as.character(V1), data = t)

However, it might be desirable to have sort() working for time/date
objects (such as POSIXct, Date and date). sort() would just have to
preserve the class as it did in R 2.2.1.
Z

> Running the code below in 2.2.1:
> V1
> 2006-08-01 2006-08-01
>                1             1
> 
> With 2.3.1 I get:
> V1
> 1154354400 1154440800
>          1          1
> 
> # testdate.R
> t <- read.csv2('testdate.csv', header=FALSE)
> t$V1 <- as.POSIXct(t$V1)
> print(t)
> x <- xtabs(V2 ~ V1, data=t)
> print(x)
> 
> # testdate.csv
> 2006-8-1;0;1
> 2006-8-1;1;1
> 2006-8-2;0;1
> 2006-8-2;0;0
> 2006-8-2;1;1
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html and provide commented,
> minimal, self-contained, reproducible code.
>