Dear all,
I would like point out a potential bug (or at least inconsistency) in the
way the R base package converts numeric numbers to dates. Specifically,
consider the following two calls:
> as.Date(0.5, origin="1969-12-31")
[1] "1970-01-01"
> as.Date(0.5, origin="1970-01-01")
[1] "1970-01-01"
Well, 2.11.0 is overdue for an update: see the posting guide.
You've confused the printed representation with the object. If there
is an issue, it is in the formatting (and hence printing) of
fractional dates before the epoch ("1970-01-01 00:00"). As the help
page says
It is intended
On 08/11/2010 3:14 PM, Dan Tenenbaum wrote:
Hello,
I think there is a problem in recent devel builds of R on Windows with
various devices from the grDevices package.
This should be fixed now, as of today's build. Thanks for the report
(and thanks to Brian Ripley for fixing it).
Duncan Murd
Hi,
Found in ?eapply (in current R-devel):
\value{
A named (unless \code{USE.NAMES = FALSE}) list. Note that the order of
the components is arbitrary for hashed environments: currently it will
be in the order the name-value pairs were added for unhashed
