R help - Apr 2007 - commandArgs usage and --args invokation

Dear R experts:

I am a bit stymied by how the argument picking-off works in R batch file usage.

$ cat commandArgs.R
cat(" Command Line Arguments were ", commandArgs(), "\n");

$ /usr/bin/R CMD BATCH commandArgs.R --args 1 2 3

$ /usr/bin/R --args 1 CMD BATCH commandArgs.R
... I am now getting into interactive mode ?!  I guess it really is skipping the
rest of the command line

$ /usr/bin/R CMD BATCH commandArgs.R --args 1

$ ls -lt | head
total 1628148
-rw-rw-r-- 1 ivo ivo       827 Mar 31 21:47 --args
...


Could someone please let me know?

[also, are there multi-core versions of R?  my guess is no, but since
I have a second processor lying around, I am trying to get it into the
fray.]

A working example would be appreciated.

regards,

/ivo welch

On Sat, 31 Mar 2007, ivo welch wrote:
> Dear R experts:
>
> I am a bit stymied by how the argument picking-off works in R batch file 
> usage.
You seem to mean 'R CMD BATCH' usage (there are other ways to run R in 
batch), and also I will assume you are talking only about a Unix-alike 
(these things are somewhat different on Windows).  Please do follow the R 
posting guide and provide basic information (and read the help).

In R 2.5.0 you probably want to use 'Rscript' instead for this sort of 
batch usage.
> $ cat commandArgs.R
> cat(" Command Line Arguments were ", commandArgs(),
"\n");
We don't want the trailing ';': this is R not C.
> $ /usr/bin/R CMD BATCH commandArgs.R --args 1 2 3
>
> $ /usr/bin/R --args 1 CMD BATCH commandArgs.R
> ... I am now getting into interactive mode ?!  I guess it really is 
> skipping the rest of the command line
Yes, that is what --args is documented to do, skip the rest of the command 
line.  Try R --help
> $ /usr/bin/R CMD BATCH commandArgs.R --args 1
>
> $ ls -lt | head
> total 1628148
> -rw-rw-r-- 1 ivo ivo       827 Mar 31 21:47 --args
> ...
Please DO read the help here:

gannet% R CMD BATCH --help
Usage: R CMD BATCH [options] infile [outfile]

so you have 'options' in the wrong place, and --args is matching 
'outfile'.  As is says that only arguments starting with '-' are
considered to be options, you cannot use '1' here as an option.  What
you
can do is

R CMD BATCH '--args 1' commandArgs.R

Once again, this is doing what it is documented to do.

> Could someone please let me know?
>
> [also, are there multi-core versions of R?  my guess is no, but since
> I have a second processor lying around, I am trying to get it into the
> fray.]
>
> A working example would be appreciated.
R is not multithreaded, but you can make use of multithreaded BLASes: see 
the R-admin installation manual.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Ivo,

For R 'shell scripts', esp. on Unixy system (incl OS X), our littler
frontend
can be of help. See

    http://dirk.eddelbuettel.com/code/littler.html

or

    http://biostat.mc.vanderbilt.edu/LittleR

as well as the SVN archive

    svn checkout http://littler.googlecode.com/svn/trunk/ littler

Using littler scripts, you can then do things like

    edd at basebud:~/src/debian/Misc/svn/littler> head -20 examples/pace.r
    #!/usr/bin/env r
    #
    # a simple example to convert miles and times into a pace
    # where the convention is that we write e.g. 37 min 15 secs
    # as 37.15 -- so a call 'pace.r 4.5 37.15' yields a pace of
    # 8.1667, ie 8 mins 16.67 secs per mile
    
    if (is.null(argv) | length(argv)!=2) {
    
      cat("Usage: pace.r miles time\n")
      q()
    
    }
    
    dig <- 5
    
    rundist <- as.numeric(argv[1])
    runtime <- as.numeric(argv[2])

    [...]

which, while not as direct as getopt in bash or C, allows you to pick
arguments off the command-line. E.g.

    edd at basebud:~> pace.r 3 25
    Miles   :  3
    Time    :  25
    Pace    :  8 min 20 sec
    edd at basebud:~>


I use littler as a command-line frontend to most of my (batch) R work for
simulation or recurring data analysis.

As Prof Ripley noted. R 2.5.0 will give you something similar via Rscript.

Hth, Dirk




-- 
Hell, there are no rules here - we're trying to accomplish something. 
                                                  -- Thomas A. Edison