[R] local source packages install from within R session - cross-platform
Ken Termiso
jerk_alert at hotmail.com
Fri Dec 9 21:18:01 CET 2005
I realize that others have struggled with this issue...i.e.
http://tolstoy.newcastle.edu.au/~rking/R/help/05/01/9826.html
i am on os.x 10.4 w/ R2.2, and am (perhaps foolishly) also on this quest...
i would like to be able to install downloaded source (tar.gz'd) files from
within an R session, and have it work in a X-platform way..i am often not
connected to the internet and have libraries that i would like to be able to
install in an automated way..
i've tried several things:
1)
install.packages("hs95av2hs3ug6cdf_6.0.0.tar.gz", repos=NULL,
method="source")
Error in gzfile(file, "r") : unable to open connection
In addition: Warning message:
cannot open compressed file 'hs95av2hs3ug6cdf_6.0.0.tar.gz/DESCRIPTION'
>traceback()
5: gzfile(file, "r")
4: read.dcf(file.path(curPkg, "DESCRIPTION"), c("Package", "Version"))
3: unpackPkg(pkgs[i], pkgnames[i], lib, installWithVers)
2: .install.macbinary(pkgs = pkgs, lib = lib, contriburl = contriburl,
method = method, available = available, destdir = destdir,
installWithVers = installWithVers, dependencies = dependencies)
1: install.packages("hs95av2hs3ug6cdf_6.0.0.tar.gz", repos = NULL,
method = "source")
2) the code in the above thread doesn't work for me on OS X..
file.pkg <- "hs95av2hs3ug6cdf_6.0.0.tar.gz"
path.pkg <- file.path(getwd(), file.pkg)
pkg_path <- paste("file://", path.pkg, sep = "")
install.packages("hs95av2hs3ug6cdf_6.0.0", contriburl=pkg_path, lib =
Sys.getenv("R_LIBS"))
Error in gzfile(file, "r") : unable to open connection
In addition: Warning message:
cannot open compressed file
'/Users/ken/Desktop/hs95av2hs3ug6cdf_6.0.0.tar.gz/PACKAGES'
>traceback()
5: gzfile(file, "r")
4: read.dcf(file = tmpf, fields = flds)
3: available.packages(contriburl = contriburl, method = method)
2: .install.macbinary(pkgs = pkgs, lib = lib, contriburl = contriburl,
method = method, available = available, destdir = destdir,
installWithVers = installWithVers, dependencies = dependencies)
1: install.packages("hs95av2hs3ug6cdf_6.0.0", contriburl = pkg_path,
lib = Sys.getenv("R_LIBS"))
*with both of those, things get screwy with gzfile(), but calling it
directly on the file is fine..
>gzfile("hs95av2hs3ug6cdf_6.0.0.tar.gz")
description class
mode
"hs95av2hs3ug6cdf_6.0.0.tar.gz" "gzfile"
"rb6"
text opened
can read
"text" "closed"
"yes"
can write
"yes"
3) i tinkered with the installLocalPackages() from sciViews library...
"installLocalPackages" <-
function(basedir = "d:/R", url = paste("file:",
basedir, "/bin/windows/contrib", sep = "")) {
# Install one or several packages from a local source
if (.Platform$OS.type != "windows")
stop("This function is for Windows only! Use install.packages instead.")
localurl <- (length(grep("^file:", url)) > 0)
if (!localurl)
stop("installLocalPackages can only install local packages. Use
install.packages to install accross the internet")
a <- CRAN.packages(contriburl = url)
pkgs <- select.list(a[, 1], , TRUE)
if (length(pkgs) == 0 || pkgs[1] == "") # The user cancelled the list
return(invisible(NULL))
lib <- .libPaths()[1]
# Install those selected packages
install.packages(pkgs, lib, CRAN = NULL)
invisible()
}
this was my final version before i quit -
"installLocalPackages" <-
function(url = paste("file:/",
basedir, sep = ""), basedir = file.path(getwd()), localPkg) {
# Install one or several packages from a local source
localurl <- (length(grep("^file:", url)) > 0)
if (!localurl)
stop("installLocalPackages can only install local packages. Use
install.packages to install accross the internet")
# a <- available.packages(contriburl = url)
# pkgs <- select.list(a[, 1], , TRUE)
# if (length(pkgs) == 0 || pkgs[1] == "") # The user cancelled the list
# return(invisible(NULL))
lib <- .libPaths()[1]
# Install those selected packages
install.packages(localPkg, lib, CRAN = NULL)
invisible()
}
ideally, i would like to have installLocalPackages() work in a X-platform
way with NO gui; i would like it to accept either a single .tar.gz or
several at a time...
any help would be much appreciated - i'm out of ideas...
More information about the R-help
mailing list