[R] s3 methods on S4 objects
Martin Morgan
mtmorgan at fhcrc.org
Thu Jan 13 19:33:33 CET 2011
On 01/13/2011 09:49 AM, steven mosher wrote:
>
>
> On Thu, Jan 13, 2011 at 6:46 AM, Martin Morgan <mtmorgan at fhcrc.org
> <mailto:mtmorgan at fhcrc.org>> wrote:
>
> On 01/12/2011 10:54 PM, steven mosher wrote:
> > I have J Chambers wonderful text ( Software for data Analysis) and
> I've been
> > trying
> > my hand at some very routine S4 OOP development.
> >
> > One of the things I was trying to do was to create some very basic S4
> > classes. The first
> > was simply a class that had a data.frame as part of its
> representation.
> >
> > setClass("df",representation(dirframe="data.frame"))
> >
> > The object basically contains a data.frame that represents a file
> directory
> > listing
> > with a column named filename, size, time, etc.
> >
> > And then I have methods for doing various things with this object.
> >
> > I then tried to tackle the problem of coercing this S4 object to a
> > data.frame. Again just a learning exercise.
> >
> > The goal would be able to make a call like this
> >
> > testFrame <- as.data.frame(x)
> >
> > where x, was an object of class "df"
> >
> > If I try to define "as.data.frame" as a S4 method, then I can make
> it work,
> > but I then destroy the S3 functionality
> > of as.data.frame, so that if I were to try to coerce a matrix to a
> > data.frame it would work.
>
> Hi Steven --
>
> This works for me
>
> setClass("A", representation=representation(df="data.frame"))
>
> setMethod("as.data.frame", "A",
> function(x, row.names=NULL, optional=FALSE, ...)
> {
> ## implementation, e.g.,
> callGeneric(x at df, row.names=row.names, optional=optional, ...)
> })
>
>
> this makes no sense to me.
>
> Looking at this in the manual:
> "A call to callGeneric can only appear inside a method definition. It
> then results in a call to the current generic function. The value of
> that call is the value of callGeneric. While it can be called from any
> method, it is useful and typically used in methods for group generic
> functions."
callGeneric is not necessary, and I should just have said
as.data.frame(x at df, row.names=row.names, optional=optional, ...)
> I'm further confused. what is the "current" generic function?
I think of a generic 'foo' as a function, and inside that function are
functions foo,A-method, foo,B-method etc for classes A, B, .... When in
one of these methods, foo,A-method, the current generic is the function
'foo'. For as.data.frame, the generic is unambiguous ('as.data.frame' !)
and the interesting case are the group generics (?Logic, for instance),
where a single method
setMethod("Logic", function(e1, e2) callGeneric(<YOUR CODE HERE>))
and you'll have in effect written 'methods' for all the operators
defined in the 'Logic' group. Cool.
>
> > as.data.frame(new("A"))
> Object of class "data.frame"
> data frame with 0 columns and 0 rows
> > as.data.frame(matrix(0, 3, 5))
> V1 V2 V3 V4 V5
> 1 0 0 0 0 0
> 2 0 0 0 0 0
> 3 0 0 0 0 0
>
>
> Maybe you call setGeneric (no need to, setMethod will promote
> as.data.frame automatically) in a way that does not specify the default
> (arg useAsDefault) correctly?
>
>
> I think that may have been the mistake.. what do you mean by
> no need to call setGeneric?
The only code I had was what was written above -- setClass and
setMethod. I could also have
setGeneric("as.data.frame")
and would have been ok -- the default behavior of setGeneric in this
case is to make a generic function as.data.frame, AND a method
as.data.frame,ANY-method. The as.data.frame,ANY-method is implemented as
base::as.data.frame, and is where objects not handled by methods I
implement might end up being dispatched to. In a new R session, try
setGeneric("as.data.frame")
showMethods(as.data.frame)
selectMethod(as.data.frame, "ANY")
If I had done
> setGeneric("as.data.frame",
function(x) standardGeneric("as.data.frame"))
Creating a generic for 'as.data.frame' in package '.GlobalEnv'
(the supplied definition differs from and overrides the implicit generic
in package 'base': Formal arguments differ: (x), (x, row.names,
optional, ...))
[1] "as.data.frame"
then I'm in trouble -- I've created a generic 'as.data.frame', but since
the signature of my generic differs from the signature of
base::as.data.frame, the default behavior does NOT create a
as.data.frame,ANY method.
Not sure if this helps or not...
Martin
>
> I really like chambers book, but there are certain parts where the lack
> of simple examples
> really makes it difficult to follow.
>
>
> Martin
>
>
> >
> >
> > So, I guess my question is what do I do, write an s3 method for
> > as.data.frame that takes a "df" object as a paramter?
> > The book wasn't exactly clear ( or I'm not that bright), or is
> there a way
> > to make the S4 method I wrote "as.data.frame"
> > call the S3 method if needed?
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help at r-project.org <mailto:R-help at r-project.org> 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.
>
>
> --
> Computational Biology
> Fred Hutchinson Cancer Research Center
> 1100 Fairview Ave. N. PO Box 19024 Seattle, WA 98109
>
> Location: M1-B861
> Telephone: 206 667-2793
>
>
--
Computational Biology
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N. PO Box 19024 Seattle, WA 98109
Location: M1-B861
Telephone: 206 667-2793
More information about the R-help
mailing list