[R] Installing local package under Windows.

Rolf Turner rolf at math.unb.ca
Fri Oct 11 20:00:42 CEST 2002

I'm working on a system in which PCs (running various versions
of Windows, mainly 98 I think) are networked together using
a system called ``Novell''.

I recently got our Computing Services people to install R
on this system, and it appears to work seamlessly.

I mainly want to use R in a time series course which I am
teaching this term, and to that end I wanted to install a
supplementary package of time series utilities which I
have written, for my students' benefit.

I cannot get this locally built package to install under
this Windows system.  (The package was built under Unix, and
installs there, with no complaints.)

I have been trying to follow the instructions in the ``R Installation
and Administration'' guide, section 5.1 ``Installing packages.''

I proceeded as follows:
1. Uploaded the zip file ``ts.sup_0.0-0.tar.gz'' from the
   Unix box to the Windows system, via ftp (binary mode set).

2. Placed the zip file in F:\rproject

	--- the F: drive is effectively one's ``login directory''
	when one has connected to the Novell network.
	--- the directory ``rproject'' seems to be created by R
	as the base location in which to do its thing.

3. Executed

	> .libPaths("L:/statdata")

   to make ``L:/statdata'' the first entry in .libPaths().

   I have write permission in L:/statdata (and of course not
   elsewhere on the network drives).  This --- i.e. L:/statdata ---
   is the location where our stats group routinely places data sets
   for the students to access.

   Having done this I executed


   to see if it had ``taken'' and got

   [1] "L:/statdata"                "O:/EXEFILES/RW1051/library"

   as hoped.  (The second entry is the default value of .libPaths().)

4. Clicked on the Packages button of the R gui, and then on

   Install package from local zip file...

5. An application box popped up as a result; I had to type in
   the name of the zip file since the point-and-click seemed to
   insist that the extension be ``.zip'' (whereas it was/is actually

The following set of error messages resulted:

> install.packages("F:/rproject/ts.sup_0.0-0.tar.gz", .libPaths()[1], CRAN = NULL)
updating HTML package descriptions
Error in file(f.tg, open = "w") : unable to open connection
In addition: Warning messages: 
1: error 1 in extracting from zip file 
2: cannot update HTML package index in: make.packages.html(lib.loc) 
3: cannot open file `O:\EXEFILES\RW1051/doc/html/search/index.txt' 

So it can't open some ``connection'', can't do a bunch of other
things it thinks it ought to do, and seems to be looking at the
O: drive rather than the L: drive to which I directed its attention.

Can anyone suggest anything I might do to remedy the situation?

Do I need to process the zip file in a different way?  I tried
changing the extension to .zip, but of course that didn't help.

Is some piece of software missing?

If nothing can be done about the ``install' problem, is there any way
--- instead of installing a package --- that I can make the contents
of ``ts.sup'' available to my students over the Novell network.  (The
contents consist of functions, data sets, and documentation thereon.)

I'm groping around pretty blindly; I have no feel for Windows at all,
and our Computing Services people who run the Novell network know
nothing about R, so I'm unlikely to get any help from them.

Any advice greatly appreciated.


				Rolf Turner
				rolf at math.unb.ca

P. S. Version details:

> version
platform i386-pc-mingw32
arch     i386           
os       mingw32        
system   i386, mingw32  
major    1              
minor    5.1            
year     2002           
month    06             
day      17             
language R 

(I wonder a bit about that ``mingw32'' --- Windows 3.2??? ---
but that's what it says.)

				R. T.
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list