CRAN packages filehash, hash, timeSeries and trip
Prof Brian Ripley
ripley at stats.ox.ac.uk
Sat Dec 5 08:56:23 CET 2009
John Chambers added the following in r50609 in R-devel:
o The R grammar prevents using the argument "name" in signatures
of methods for "$" and "$<-". The generic functions have been
changed to reflect this.
(He means the implicit S4 generics: the actual functions are
unchanged.)
It has had to be temporarily reverted, but will be restored soon.
This means that packages that define methods for $ or $<- with 'name'
in the signature (which was pointless as it is always class
"character") need to be corrected: these were sp, rgdal and those
listed in the subject.
For setMethod("$" ...) all you have ever needed was signature
"myclass". In principle for setReplaceMethod("$" ...) you could give
a signature of the form signature(x="myclass", value="someclass") but
as implemented that does not work: only the class of 'x' is used in
dispatching. So in both cases the signature should simply be the
class of x.
Can we please have updates that do this.
--
Brian D. Ripley, ripley at stats.ox.ac.uk
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UK Fax: +44 1865 272595
More information about the Rmetrics-core
mailing list