Herve Pages
2007-Jan-29 23:02 UTC
[Rd] update.packages() doesn't pick up the highest version of packages
Hi, Try to install limma 2.8.1 with: > biocRep <- "http://bioconductor.org/packages/1.9/bioc" > install.packages("limma", repos=biocRep) then try to update it with > repos <- c(biocRep, "http://cran.fhcrc.org") > update.packages(repos=repos) --> it will not get updated (even if there is a 2.9.8 version on CRAN). This is (1) surprising by itself, (2) inconsistent with install.packages behaviour. When given more than 1 repository, install.packages() will go thru all of them and pick up the higher version of each package to install. For example, this: > repos <- c(biocRep, "http://cran.fhcrc.org") > install.packages("limma", repos=repos) will install limma from CRAN (2.9.8) instead of limma from Bioconductor (2.8.1). But if a new version of limma shows up on CRAN, this: > update.packages(repos=repos) will not update it... Cheers, H.
Has anyone ever suggested making segments() and arrows() generic (as indeed they are in another not dissimilar implementation)? The practical issue with these functions is the need to specify origin and destination using four separate arguments. It would be much more convenient to use only two. If segments were (S4) generic, for example, you could write methods for signature(x0 = "matrix", y0 = "matrix", x1 = "missing", y1 = "missing") signature(x0 = "complex", y0 = "complex", x1 = "missing", y1 "missing") With the present function as the default method, no existing working code would be affected. Bill Venables