[R] Is this foreach behaviour correct?

[Henrik Bengtsson]

It looks like a bug.  I don't think c.Date() is every called, because:

> trace(c.Date, tracer = quote(message("c.Date() called")))
Tracing function "c.Date" in package "base"
[1] "c.Date"

Tracing works:

> c(as.Date(10000L), as.Date(10001L))
Tracing c.Date(as.Date(10000L), as.Date(10001L)) on entry
c.Date() called
[1] "1997-05-19" "1997-05-20"

but c.Date() is not called here:

> x <- foreach(i=10000:10100, .combine = function(...) c(...)) %do% { as.Date(i) }
> str(x)
 num [1:101] 10000 10001 10002 10003 10004 ...


The following hack works:

> library("foreach")
> library("zoo")
> x <- foreach(i=10000:10100, .combine = function(...) c(...)) %do% { as.Date(i) }
> str(x)
 Date[1:101], format: "1997-05-19" "1997-05-20" "1997-05-21" "1997-05-22" ...

Alternatively, one can use append() which works like c() if no other
arguments are specified:

> x <- foreach(i=10000:10100, .combine = append) %do% { as.Date(i) }
> str(x)
 Date[1:101], format: "1997-05-19" "1997-05-20" "1997-05-21" "1997-05-22" ...

It looks like foreach is treating the .combine = c case specially and
someone fail to properly dispatch c() on the object (or something).

/Henrik


On Sun, Nov 13, 2016 at 7:14 AM, James Hirschorn
<james.hirschorn at hotmail.com> wrote:
> I'm still not clear about whether this is a bug in foreach. Should c.Date be invoked by foreach with .combine='c'?
>
> On 11/06/2016 07:02 PM, William Dunlap wrote:
> Note that in the OP's example c.Date is never invoked.  c.Date is called if .combine
> calls c rather than if .combine is c:
>
>> library(zoo)
>> trace(c.Date, quote(print(sys.call())))
> Tracing function "c.Date" in package "base"
> [1] "c.Date"
>> foreach(i=10000:10003, .combine=c) %do% { as.Date(i) }
> [1] 10000 10001 10002 10003
>> foreach(i=10000:10003, .combine=function(...)c(...)) %do% { as.Date(i) }
> Tracing c.Date(...) on entry
> eval(expr, envir, enclos)
> Tracing c.Date(...) on entry
> eval(expr, envir, enclos)
> Tracing c.Date(...) on entry
> eval(expr, envir, enclos)
> [1] "1997-05-19" "1997-05-20" "1997-05-21" "1997-05-22"
>
>
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com<http://tibco.com>
>
> On Sun, Nov 6, 2016 at 2:20 PM, Duncan Murdoch <murdoch.duncan at gmail.com<mailto:murdoch.duncan at gmail.com>> wrote:
> On 06/11/2016 5:02 PM, Jim Lemon wrote:
> hi James,
> I think you have to have a starting date ("origin") for as.Date to
> convert numbers to dates.
>
> That's true with the function in the base package, but the zoo package also has an as.Date() function, which defaults the origin to "1970-01-01".  If James is using zoo his code would be okay.  If he's not, he would have got an error, so I think he must have been.
>
> Duncan Murdoch
>
>
>
> Jim
>
> On Sun, Nov 6, 2016 at 12:10 PM, James Hirschorn
> <james.hirschorn at hotmail.com<mailto:james.hirschorn at hotmail.com>> wrote:
> This seemed odd so I wanted to check:
>
>  > x <- foreach(i=10000:10100, .combine='c') %do% { as.Date(i) }
>
> yields a numeric vector for x:
>
>  > class(x)
> [1] "numeric"
>
> Should it not be a vector of Date?
>
> ______________________________________________
> R-help at r-project.org<mailto:R-help at r-project.org> mailing list -- To UNSUBSCRIBE and more, see
> 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.
>
> ______________________________________________
> R-help at r-project.org<mailto:R-help at r-project.org> mailing list -- To UNSUBSCRIBE and more, see
> 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.
>
>
> ______________________________________________
> R-help at r-project.org<mailto:R-help at r-project.org> mailing list -- To UNSUBSCRIBE and more, see
> 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.
>
>
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.