R "sumo" package suggestion

Hello,

I think Achim suggestion is more realistic: it does not imply automatic
installation of all packages, but just a restricted list of packages
available on CRAN about a specific topic.

An easy way to get this result is to propose separate lists of  
packages. It
means separate lists than
http://cran.r-project.org/src/contrib/PACKAGES.html (and 'PACKAGES' in  
the
Windows packages binaries). I don't see the problem to propose other  
lists
that could be called 'SpatialStats.html' (in packages
sources)/'SpatialStats' (in Windows binaries)... and the same for
'MachineLearning', 'Biostats', etc...

Then, of course the various functions that install packages should be
adapted to use these lists. It does not look like an unsurmontable  
task.

Of course, if this is not done yet by the R Core Team, I presume that  
there
must be difficulties that I don't see. It is obvious that, either we  
need a
list maintainer for each topic, or we have to propose keywords for  
packages
(similar to the keywords for functions) that will be used to  
automatically
generate those separate lists.

An alternative that can currently be used for groups of users in an
institution is to maintain a local copy of R packages repository, which
contains only the packages of interest for this group. I do so for my
students. Under Windows, in the new R 2.0.1 beta, there is a new menu  
entry
in packages -> Set CRAN mirror... (in my version it does not work yet,
looking for a missing .\doc\CRAN_mirrors.csv file), but I can easily  
figure
out how it works and how I could append my own repository to the list  
to
ease installation of a restricted list of R packages by my students.  
This is
only for Windows, but a similar approach can also be used on other  
platforms
with a little bit of coding.

Best,

Philippe

..............................................<??}))><........
 ) ) ) ) )
( ( ( ( (    Prof. Philippe Grosjean
 ) ) ) ) )
( ( ( ( (    Numerical Ecology of Aquatic Systems
 ) ) ) ) )   Mons-Hainaut University, Pentagone
( ( ( ( (    Academie Universitaire Wallonie-Bruxelles
 ) ) ) ) )   6, av du Champ de Mars, 7000 Mons, Belgium
( ( ( ( (
 ) ) ) ) )   phone: + 32.65.37.34.97, fax: + 32.65.37.33.12
( ( ( ( (    email: Philippe.Grosjean at umh.ac.be
 ) ) ) ) )
( ( ( ( (    web:   http://www.umh.ac.be/~econum
 ) ) ) ) )
..............................................................

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Achim Zeileis
Sent: Thursday, November 11, 2004 10:21 PM
To: Liaw, Andy
Cc: r-help at stat.math.ethz.ch; rsparapa at post.its.mcw.edu
Subject: Re: [R] R "sumo" package suggestion

On Thu, 11 Nov 2004 15:59:09 -0500 Liaw, Andy wrote:

Good idea, IMHO, but there are some practical difficulties:

I guess the XEmacs packages are (most, if not all) pure elisp code,
and do not need other stuff to work.  However, quite a few CRAN
packages depend on external libraries or programs, and do not
necessarily work on all platforms that R runs on.  How would such
dependencies be resolved in such a kitchen sink bundle?

I have a somewhat related idea:  Start labelling packages

with a set

of pre-defined categories, and a package can be labelled with more
than one categories (especially those *misc type packages).  It is
then possible to have facility to let people install all

packages that

fall in a particular category (e.g., `spatial statistics').

 I believe

several systems have such facilities, Debian being one of them,
TeXLive being another.

This is similar to idea that has been discussed from time to
time for several years now: it would be nice to have
maintained "CRAN task views"
(or something like that), i.e., we could have a maintainer
for, say "spatial stats", another one for "machine learning",
"biostats" which can of course be overlapping. Then the
maintainers would have to produce some sort of list of
packages (in a standardized format) with a little bit of
markup such that a web page can be generated from it and that
the information could be used by install.packages().
I think most users would profit from that, but nobody has
done the work to provide the infrastructure so far. I've just
discussed this with Kurt again, a week ago or so...I wanted
to play around with some ideas, but didn't get round to
really do something yet. But hopefully, I'll get round to
work on this in the next weeks.
Z

Just my $0.02...

Andy

From: Rodney Sparapani

r-help:

I have an R package suggestion.  After spending several hours the
other day installing about a dozen packages, I had an idea.  In
xemacs, there is a "sumo" package which allows me to

install a large

bundle of xemacs packages at one time (about a 120 modes

including

ESS).  I think R should have a similar bundle.  It would

be so much

easier than hunting/downloading/installing.  Martin

encouraged me to

send this suggestion to r-help.  In addition, he put

together a few

comments relating to the previous times that this, or a similar
suggestion, has been brought up here.

Martin wrote:

If you search for "install all CRAN packages"
on
	http://maths.newcastle.edu.au/~rking/R/

(the URL which is quickly found from the [Search] sidebar of
http://www.R-project.org/)

You find things like Greg Warnes 'Makefile'
http://tolstoy.newcastle.edu.au/R/help/04/04/0723.html
and
http://tolstoy.newcastle.edu.au/R/help/04/04/0616.html
which is from Tony and has the following small function:

  installNewCRANPackages <- function() {
    ## (C) A.J. Rossini, 2002--2004
    test2 <- packageStatus()$avail["Status"]

install.packages(row.names(test2)[which(test2$Status=="not
installed")])
  }

----------

Rodney Sparapani              Medical College of Wisconsin
Sr. Biostatistician           Patient Care & Outcomes Research
rsparapa at mcw.edu              http://www.mcw.edu/pcor
Was 'Name That Tune' rigged?  WWLD -- What Would Lombardi Do