[Rd] Problem with initialize of S4 classes

[Martin Morgan]

Hi Christian --

Does your package have a name space, but not export 'initialize'?
Calling 'new' will then go directly to the default constructors,
generating the error. A different solution would be to define a
constructor in your package

PreFilter <- function(...) {
    new("PreFilter", ...)
}

so that your user would not have to know about the details of calling
new, you could provide helpful arguments to your constructor, and
'pre-processing' of arguments can be done (in the constructor) before
calling 'new' (I find this helpful, to avoid have to be too careful
when constructing initialize methods that deal with both the class and
classes derived from the class).

Martin

cstrato <cstrato at aon.at> writes:

> Dear all
>
> Below is the code for "scriptPreFilter.R" which gives the following result:
>  > source("scriptPreFilter.R")
>  > prefltr <- new("PreFilter", mad=c(0.5,0.01))
> [1] "------initialize:PreFilter------"
> [1] "------initialize:Filter------"
>  > str(prefltr)
> Formal class 'PreFilter' [package ".GlobalEnv"] with 2 slots
>   ..@ mad       :List of 2
>   .. ..$ cutoff : num 0.5
>   .. ..$ epsilon: num 0.01
>   ..@ numfilters: num 1
>
> It seems that everything is ok, the results are as expected.
>
> However, my problem is that copying the identical code to my package 
> results in an error:
>  > prefltr <- new("PreFilter", mad=c(0.5,0.01))
> [1] "------setValidity:Filter------"
> Error in validObject(.Object) :
>   invalid class "PreFilter" object: invalid object for slot "mad" in 
> class "PreFilter": got class "numeric", should be or extend class "list"
>
> The following code avoids the error and gives the result:
>  > prefltr <- new("PreFilter", mad=list(0.5,0.01))
> [1] "------setValidity:Filter------"
> [1] "------setValidity:PreFilter------"
>  > str(prefltr)
> Formal class 'PreFilter' [package "xps"] with 11 slots
>   ..@ mad        :List of 2
>   .. ..$ : num 0.5
>   .. ..$ : num 0.01
>   ..@ numfilters : num 0
>
> This is only partly correct, e.g. numfilters is 0.
>
> Only the following code gives the correct result:
>  > prefltr <- new("PreFilter")
>  > madFilter(prefltr) <- c(0.5,0.01)
>  > str(prefltr)
> Formal class 'PreFilter' [package "xps"] with 11 slots
>   ..@ mad        :List of 2
>   .. ..$ cutoff : num 0.5
>   .. ..$ epsilon: num 0.01
>   ..@ numfilters : num 1
>
> As you see, the loading "scriptPreFilter.R" calls method initialize but 
> not setValidity.
> In contrast, loading my package as library calls setValidity but not 
> initialize.
>
> My question is:
> - Why can the identical code behave differently when put in a package?
> - How can I ensure, that initialize gets also called in my package?
>
>  > sessionInfo()
> R version 2.6.1 (2007-11-26)
> i386-apple-darwin8.10.1
>
> locale:
> C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base     
>
> other attached packages:
> [1] xps_0.4.0
>
> loaded via a namespace (and not attached):
> [1] rcompgen_0.1-17
>
> Best regards and Merry Christmas
> Christian
> _._._._._._._._._._._._._._._._
> C.h.i.s.t.i.a.n S.t.r.a.t.o.w.a
> V.i.e.n.n.a       A.u.s.t.r.i.a
> e.m.a.i.l:    cstrato at aon.at
> _._._._._._._._._._._._._._._._
>
>
> ------- BEGIN: scriptPreFilter.R ---------
> setClass("Filter",
>    representation(numfilters = "numeric"),
>    prototype(numfilters = 0)
> )
>
> setClass("PreFilter",
>    representation(mad = "list"),
>    contains=c("Filter"),
>    prototype(mad = list())
> )
>
> setGeneric("madFilter",   function(object) standardGeneric("madFilter"))
> setGeneric("madFilter<-", function(object, value) 
> standardGeneric("madFilter<-"))
>
> "initialize.Filter" <- function(.Object, ...)
> {
>    print("------initialize:Filter------")
>    .Object <- callNextMethod(.Object, ...)
>    .Object
> }
> setMethod("initialize", "Filter", initialize.Filter)
>
> setValidity("Filter",
>    function(object) {
>       print("------setValidity:Filter------")
>       msg <- NULL
>       if (is.null(msg)) TRUE else msg
>    }
> )
>
> "initialize.PreFilter" <- function(.Object, mad = list(), ...)
> {
>    print("------initialize:PreFilter------")
>    .Object at numfilters <- 0
>    if (length(mad)) madFilter(.Object) <- unlist(mad)
>    .Object <- callNextMethod(.Object, ...)
>    .Object
> }
> setMethod("initialize", "PreFilter", initialize.PreFilter)
>
> setValidity("PreFilter",
>    function(object) {
>       print("------setValidity:PreFilter------")
>       msg <- NULL
>       if (is.null(msg)) TRUE else msg
>    }
> )
>
> setMethod("madFilter", signature(object="PreFilter"),
>    function(object) object at mad
> )
>
> setReplaceMethod("madFilter", signature(object="PreFilter", 
> value="numeric"),
>    function(object, value) {
>       if (length(value) == 1) {
>          value[2] <- 0.01
>       } else if (length(value) != 2) {
>          stop(paste(sQuote("mad"), "must have <cutoff,epsilon>"))
>       }#if
>
>       if (length(object at mad) == 0) {
>          object at numfilters <- object at numfilters + 1
>       }#if
>       object at mad <- list(cutoff  = as.double(value[1]),
>                          epsilon = as.double(value[2]))
>       return(object)
>    }
> )
> ------- END: scriptPreFilter.R ---------
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel

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

Location: Arnold Building M2 B169
Phone: (206) 667-2793