Re: [Rd] method using several (and different) arguments in turn

From: <>
Date: Tue, 14 Feb 2012 17:57:14 +0000

Dear Martin,
I could not entirely follow your suggestion. I can see how you define two classes inheriting from TypeOfSample, and how these two classes have two associated methods (incidentally, I'm unfamiliar with the:

           function(type,x=numeric(),...)to do.


expression. What is type?

But then I'm lost. I wouldn't know how to carry on from here.

I don't know if I have explained the key point clear enough before. For the person using "newsample" it makes sense which name (x or y or z, etc) is used. newsample(x=12.4) would give something different from newsample(y=12.4); and yet another result would be obtained if using a combination, like newsample(x=12.4,y=12.4).

I wanted to use a simple function with default values at first, but I'm in the middle of developing a package using S4 formalism. I'm not sure this would be a wise thing

Sent: 14 February 2012 17:00
To: Foadi, James (Imperial Coll.,RAL,DIA) Cc:
Subject: Re: [Rd] method using several (and different) arguments in turn

On 02/14/2012 08:43 AM, wrote:
> Dear R-developers community, I have the following generic:
> setGeneric(
> name="newsample",
> def=function(x,y,z,a,b,c,...){standardGeneric("newsample")}
> And I can build several methods for this generic. One useful thing is to use "newsample"
> with only one of the 6 arguments listed. At the moment this is what I do:
> setMethod(
> f="newsample",
> signature=c("missing","missing","numeric","missing","missing","missing"),
> function(x,y,z,a,b,c,...)
> {
> ..............................
> ..............................
> }
> )
> This would be used when the single argument is z:
> newsample(z=12.5)
> To use newsample with another argument (say x) I should implement the same as before,
> but with signature c("numeric","missing","missing","missing","missing","missing").
> Is there another shorter and easier way to do this?

Hi James --

A matter of opinion, but multiple dispatch like this can be very complicated, e.g., figuring out the 'next' method when dispatching on two or more arguments; I'd really discourage it.

A different approach, assuming that x, y, z, ... are all numeric() but that the sample to be drawn differs, is to define a small class hierarchy to be used for dispatch.


   setClass("XSample", contains="TypeOfSample")    XSample <- new("XSample") ## a 'singleton', used for dispatch

   setClass("YSample", contains="TypeOfSample")    YSample <- new("YSample")

and then


       function(type, x=numeric(), ...) standardGeneric("newsample"),

   setMethod("newsample", "XSample", function(type, x=numeric(), ...) {


   setMethod("newsample", "YSample", function(type, x=numeric(), ...) {


One could implement a default method on "TypeOfSample", and use callNextMethod() after initial transformation, if that were the pattern.

To use:

   newsample(XSample, x=1:100)


> J

Computational Biology
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N. PO Box 19024 Seattle, WA 98109

Location: M1-B861
Telephone: 206 667-2793

This e-mail and any attachments may contain confidential...{{dropped:8}}

______________________________________________ mailing list
Received on Tue 14 Feb 2012 - 18:03:28 GMT

This quarter's messages: by month, or sorted: [ by date ] [ by thread ] [ by subject ] [ by author ]

All messages

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.2.0, at Wed 15 Feb 2012 - 06:20:18 GMT.

Mailing list information is available at Please read the posting guide before posting to the list.

list of date sections of archive