On Jan 29, 2010, at 16:11 PM, Michael Lawrence wrote: > c behaviour > ========= > > It seems to behave strangely in connection with do.call and a named list: > > ir = IRanges(start = 1:3, end = 2:4) > do.call(c, list(ir, ir)) ## What I expect, a single IRange with 6 entries. > do.call(c, list(ir1 = ir, ir2 = ir)) ## This I don't get > > $ir1 > IRanges of length 3 > start end width > [1] 1 2 2 > [2] 2 3 2 > [3] 3 4 2 > > $ir2 > IRanges of length 3 > start end width > [1] 1 2 2 > [2] 2 3 2 > [3] 3 4 2 > > My guess is that this is something with c and method dispatching and do.call. > > > Yes, it is known that there is strange behavior here. The combination of dispatching on a .Primitive and the "..." signature pushes things a bit too far. What I don't get is why this works when the list has no names and why it fails with a named list. This is potentially dangerous behavior. Kasper

Hi Kasper, Kasper Daniel Hansen wrote: [...] > > What I don't get is why this works when the list has no names and why it fails with a named list. This is potentially dangerous behavior. It's a bug in the dispatch code. setClass("A", representation(aa="integer")) setMethod("c", "A", function(x, ..., recursive=FALSE) { if (missing(x)) args <- unname(list(...)) else args <- unname(list(x, ...)) ans_aa <- unlist(lapply(args, function(x) x at aa)) new(class(args[[1L]]), aa=ans_aa) } ) The following works as expected (in R-2.11, not in R-2.10): > a <- new("A", aa=3:-2) > c(a, a) An object of class "A" Slot "aa": [1] 3 2 1 0 -1 -2 3 2 1 0 -1 -2 > c(x=a, a) An object of class "A" Slot "aa": [1] 3 2 1 0 -1 -2 3 2 1 0 -1 -2 > c(a1=a, a2=a) An object of class "A" Slot "aa": [1] 3 2 1 0 -1 -2 3 2 1 0 -1 -2 But when class A and its "c" method are defined in a package, then dispatch fails: > library(mypkg) > c(a1=a, a2=a) $a1 An object of class "A" Slot "aa": [1] 3 2 1 0 -1 -2 $a2 An object of class "A" Slot "aa": [1] 3 2 1 0 -1 -2 'mypkg' contains a single R file with the definitions of class A and its "c" method. It doesn't even need to have a namespace. Anybody's willing to report this to R-devel? Cheers, H. > > Kasper > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- Hervé Pagès Program in Computational Biology Division of Public Health Sciences Fred Hutchinson Cancer Research Center 1100 Fairview Ave. N, M2-B876 P.O. Box 19024 Seattle, WA 98109-1024 E-mail: hpages at fhcrc.org Phone: (206) 667-5791 Fax: (206) 667-1319