Never would have found that on my own. Could there possibly be a way to try that in the type coercion when a key gets set that is a date, time, or datetime class? I see no reason why key(x) = "date" should fail if all the things necessary to convert a date to something data.table is fine with exists.
On 26 July 2011 15:03, Matthew Dowle <[email protected]> wrote: > Hi, > How about ?IDateTime. > Matthew > > On Tue, 2011-07-26 at 14:07 -0400, Chris Neff wrote: > > Hi all, > > > > > > I often have a date index in a data frame that I don't really care to > > keep as a time series because I don't do any sort of smoothing or > > anything, but I prefer to keep it as a date class (or POSIXct or > > whatever) so I can do things like only get data past a certain date, > > or so plots will format correctly. > > > > > > I've found that any time I time to make a date class a key variable it > > complains that it can't be converted to integer without losing > > information. I've had to resort to saving them as factors and then > > using as.POSIXct whenever i want them as dates, which is annoying and > > error prone. > > > > > > Does anyone have a better way? POSIXct supports all standard equality > > functions, and (I'm pretty sure) you can just convert into unix > > seconds without losing anything, so why does it complain? > > > > > > -Chris > > _______________________________________________ > > datatable-help mailing list > > [email protected] > > > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help > > >
_______________________________________________ datatable-help mailing list [email protected] https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help
